Download MSc Thesis Report

Document related concepts

Numerical weather prediction wikipedia , lookup

Solar radiation management wikipedia , lookup

Climate sensitivity wikipedia , lookup

Atmospheric model wikipedia , lookup

Urban heat island wikipedia , lookup

Instrumental temperature record wikipedia , lookup

General circulation model wikipedia , lookup

Transcript
1
Wageningen University
MSc Thesis Report
Applicability of the Maximum Power Limit for
Estimating Land-Atmosphere Interactions on the
Diurnal Time Scales
Author:
Aljoša Slameršak
WUR number:
900829766200
Study Program: MSc Climate Studies
Supervisors:
Prof. Dr. Laurens Ganzeveld
Dr. Axel Kleidon
Chair Group:
Earth System Science
Wageningen University
Jena, 27th of April 2015
2
Abstract
Surface turbulent heat fluxes are the driving force of the land-atmosphere exchange in
the system that is in a state of a sustained thermodynamic disequilibrium. We propose
an alternative modelling approach by assuming that the atmosphere performs as a thermodynamic heat engine which is operating at its theoretical maximum power generation
limit with respect to the convective transport in the system. This assumption provides
us with an additional physical constraint that makes it possible to model the surface
turbulent heat fluxes and convective transport without making use of to the empirical
parameterizations of the systemic parameters.
The maximum power generation hypothesis is analyzed on the diurnal time frame with
an implementation of a simple modelling framework, consisting of the three vertically
coupled reservoirs. First of all is the surface reservoir which apart from absorbing and
releasing heat also regulates the surface evapotranspiration. The near surface layer of the
atmosphere is represented with an energetic boundary layer reservoir, the heart of the
atmospheric heat engine that facilitates the convective motions. The remaining top of the
atmosphere reservoir is fundamental for the longwave radiation exchange in the system.
By performing a comparative data analysis of the model outputs with the observations from a high latitude field station in Hyytiälä, we prove that the maximum power
generation limit provides reasonable first order estimates on the diurnal time scales, thus
indicating that the system indeed operates near this thermodynamic limit. This indication implies an existence of a self-regulating mechanism in the system that optimizes the
convective exchange between the surface and atmosphere accordingly with the state of
maximum power generation. We propose a qualitative reasoning for such a functioning
of the system. Furthermore, we demonstrate the added value of our simple approach in
the modelling of the land-atmosphere interactions, as it requires only a limited number of
generally accessible input variables such as solar radiation, surface temperature and water
availability. Finally, we identify the limitations of the proposed framework and outline
some improvements for the future research. We conclude by presenting the applicability of
the maximum power generation limit in the more complex models, stressing its potentials
for an improved general understanding of the land-atmosphere interactions.
i
Contents
List of Figures
1 Introduction
1.1 Applications of the Thermodynamic Limits in Research . . . . . . . . . .
1.2 Thermodynamic Limits in Earth Systems Science . . . . . . . . . . . . .
1.2.1 Thermodynamic Disequilibrium of the Earth System . . . . . . .
1.2.2 Maximum Power Generation of the Atmospheric Heat Engine . .
1.3 Energetic Boundary Layer, a Heart of the Atmospheric Heat Engine . . .
1.3.1 Energetic Boundary Layer Extension of the Two Reservoir Model
1.4 Objectives and Aims of Thesis Research . . . . . . . . . . . . . . . . . .
1.4.1 Research goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.4.2 Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 Modelling Framework
2.1 Equations of the Radiative and Convective Atmospheric Heat Engine with
the MPP Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.1 Surface Energy Balance . . . . . . . . . . . . . . . . . . . . . . .
2.1.2 Energetic Boundary Layer Energy Balance . . . . . . . . . . . . .
2.1.3 Top of the Atmosphere Energy Balance . . . . . . . . . . . . . . .
2.1.4 Power of the Atmospheric Heat Engine . . . . . . . . . . . . . . .
2.1.5 Derivation of the MPP Limit . . . . . . . . . . . . . . . . . . . .
2.2 Surface Turbulent Heat Fluxes Parametrization . . . . . . . . . . . . . .
2.3 Soil and Vegetation Modules . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.1 Soil Heat Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.2 Soil Moisture . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.2.1 Water Balance Equations in the Soil . . . . . . . . . . .
2.3.2.2 Water Availability . . . . . . . . . . . . . . . . . . . . .
2.3.3 Vegetation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.3.1 Extension of the Surface-Water Balance . . . . . . . . .
2.4 Schematic Representation of the Model . . . . . . . . . . . . . . . . . . .
iv
.
.
.
.
.
.
.
.
.
2
4
5
5
6
10
10
11
13
13
14
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
14
14
15
16
18
20
23
24
24
25
25
26
26
27
28
ii
3 Methodology
3.1 Application of the Conceptual Model . . . . . . . .
3.2 Observations . . . . . . . . . . . . . . . . . . . . . .
3.2.1 Hyytiälä Forestry Field Station Data . . . .
3.2.1.1 Soil and Vegetation Module . . . .
3.2.1.2 Solar Radiation Forcing . . . . . .
3.2.1.3 The Greenhouse Effect . . . . . . .
3.2.1.4 Surface Turbulent Heat Fluxes . .
3.2.2 Calculation of the Energetic Boundary Layer
3.2.3 Output Data Analysis . . . . . . . . . . . .
3.3 Changing Climate Sensitivity Analysis . . . . . . .
4 Results
4.1 Conceptual Model . . . . . . . . . . . . . . . . . . .
4.2 Radiative and Convective Atmospheric Heat Engine
4.2.1 Surface Energy Balance . . . . . . . . . . .
4.2.2 Surface Turbulent Heat Fluxes . . . . . . . .
4.2.2.1 H and λE Dependence on RS . . .
4.2.3 Energetic Boundary Layer Heat Storage . .
4.2.4 Vertical Exchange Velocity . . . . . . . . . .
4.2.5 Surface and Atmospheric Temperatures . . .
4.3 Changing Climate Sensitivity Analysis . . . . . . .
4.3.1 Surface Energy Balance . . . . . . . . . . .
4.3.2 Mean Surface Temperature . . . . . . . . . .
4.3.3 Surface Turbulent Heat Fluxes . . . . . . . .
4.3.4 Convective Transport . . . . . . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
Heat Storage
. . . . . . . .
. . . . . . . .
. . . .
Model
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5 Discussion
5.1 Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.1 The MPP Mechanism . . . . . . . . . . . . . . . . . .
5.1.2 Alternative Approach for the Modelling of the Surface
Fluxes . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.3 The Changing Climate Experiments . . . . . . . . . . .
5.2 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.1 The Two Reservoir Representation of the Atmosphere .
5.2.2 Soil and Vegetation Modules . . . . . . . . . . . . . . .
5.2.3 The EBL Heat Storage . . . . . . . . . . . . . . . . . .
5.2.4 Power Generation and Dissipation in a Steady State . .
5.2.5 Representation of the Water Cycle . . . . . . . . . . .
5.3 Prospective Applications of the MPP in the Complex Models .
6 Conclusions
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . . . . .
. . . . . .
Turbulent
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
.
.
.
.
.
.
.
.
.
.
29
29
30
31
31
31
32
32
32
34
35
.
.
.
.
.
.
.
.
.
.
.
.
.
37
37
40
40
41
43
44
46
48
49
49
50
51
53
55
. 55
. 55
.
.
.
.
.
.
.
.
.
58
61
63
63
64
65
66
67
68
69
iii
A Soil and Vegetation Modules
A.1 Soil Heat Flux . . . . . . . . . . . . . . . . . . . . . . . . .
A.1.1 Model Implementation . . . . . . . . . . . . . . . .
A.1.2 Impact of Moisture on the Conductive Properties of
A.1.2.1 Soil density . . . . . . . . . . . . . . . . .
A.1.2.2 Heat capacity . . . . . . . . . . . . . . . .
A.1.2.3 Thermal conductivity . . . . . . . . . . .
A.1.3 Soil Heat Flux Results . . . . . . . . . . . . . . . .
A.2 Soil Moisture . . . . . . . . . . . . . . . . . . . . . . . . .
A.2.1 Groundwater flow . . . . . . . . . . . . . . . . . . .
A.2.2 Water Availability in the Bare Soil . . . . . . . . .
A.2.3 Soil Moisture Storage Results . . . . . . . . . . . .
A.3 Vegetation . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . .
. . . . .
the Soil
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
71
71
71
72
72
73
73
74
76
76
76
77
79
B Future Recommendations
81
B.1 The EBL with Water Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . 81
B.2 The Multiple Heat Storage Extension . . . . . . . . . . . . . . . . . . . . . 83
C List of Variables and Parameters
84
References
89
iv
List of Figures
1.1
1.2
1.3
A Simple Two Reservoir Model . . . . . . . . . . . . . . . . . . . . . . . . 7
A Simple Two Reservoir Model With a Heat Engine System . . . . . . . . 9
Energetic Boundary Layer Scheme . . . . . . . . . . . . . . . . . . . . . . . 11
2.1
2.2
Illustration of the Greenhouse Recycling Mechanism . . . . . . . . . . . . . 17
A Complete Framework Scheme . . . . . . . . . . . . . . . . . . . . . . . . 28
3.1
Illustration of the EBL Sensible Heat Storage Calculation . . . . . . . . . . 33
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
4.10
The Conceptual Model Outputs . . . . . . . . . . . . . . . . . . . . . . . .
Surface Energy Balance - Measurements and the Model . . . . . . . . . . .
Sensible Heat Flux Analysis . . . . . . . . . . . . . . . . . . . . . . . . . .
Latent Heat Flux Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . .
Turbulent Heat Fluxes Dependence on Solar Radiation . . . . . . . . . . .
Energetic Boundary Layer Heat Storage Analysis . . . . . . . . . . . . . .
Average Diurnal Cylcle of the Modelled EBL Heat Storage . . . . . . . . .
Vertical Exchange Velocity . . . . . . . . . . . . . . . . . . . . . . . . . . .
Surface and Atmospheric Temperature Analysis . . . . . . . . . . . . . . .
Changes in the Surface Energy Balance - Model Scenarios: ∆L↓ = −2.8 W/m2
and ∆L↓ = 8.5 W/m2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.11 Sensitivity of the Mean Surface Temperature to a Changed Greenhouse
Forcing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.12 Sensitivity of the Surface Turbulent Heat Fluxes to a Mean Surface Temperature Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.13 Sensitivity of the Convective Velocity Scale to a Mean Surface Temperature
Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1
5.2
5.3
39
41
42
42
44
45
46
47
48
50
51
52
54
Schematic Illustration of the MPP Mechanism . . . . . . . . . . . . . . . . 57
Feedbacks in the Proposed Self-Regulating MPP Mechanism . . . . . . . . 57
Mean Characteristics of the Turbulent Heat Fluxes within the Convective
Boundary Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
A.1 Soil Heat Flux Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
v
A.2 Soil Moisture Storage Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 78
B.1 Recommended Improvements in the Framework . . . . . . . . . . . . . . . 82
vi
Acknowledgments
I express my sincere gratitude to a number of people who have contributed to this research.
First and foremost I would like to thank my supervisor Prof. Dr. Laurens Ganzeveld,
who has introduced me to the topic of thermodynamic limits in the Earth System. The
discussions we have had, his wholehearted assistance and moral support have helped me
enormously to maintain the self-confidence, perseverance and eagerness for research.
I extend my deepest thanks to my co-supervisor Dr. Axel Kleidon, whose ability to
convey the abstract thermodynamic concepts and to pose creative challenges decisively
marked the invaluable learning experience of working at the Max Planck Institute for
Biogeochemistry, Jena. I am grateful to the other colleagues from Biospheric Theory
and Modelling Group: Dr. Maik Renner, Dr. Chiraq Dhara and Dr. Lee Miller, with
whom I have been actively engaged in the conceptual design of my research. They have
all provided me with constructive feedback and research suggestions which helped me to
develop the modelling framework as well as to avoid the pitfalls on the way. I thank them
as well for kindly hosting me at their research facilities in Jena. My appreciation also
goes to Prof. Dr. Jordi Vila-Guerau de Arellano, for his helpful advice on the turbulent
heat fluxes parameterizations.
This research would not be possible without the observation data. Observation data
from the field station was a courtesy of AVAA, an open data publishing platform, initialized by the Finnish Ministry of Education and Culture. Radiosonde measurements were
a courtesy of HUMPPA/COPEC field campaign, provided by the WUR Earth System
Science Chair Group.
Lastly, I would like to thank the Slovene human resources development and scholarship
fund for their financial support during the two year of my studies at the WUR.
1
Table of Abbreviations and Acronyms
Acronym/Abbreviation
Meaning
MEP
Maximum entropy production principle
MPP
Maximum power generation principle
GCMs
General circulation models
NEE
Net entropy exchange
EBL
Energetic boundary layer
TOA
Top of the atmosphere
LAI
Leaf area index
IPCC
Intergovernmental Panel on Climate Change
EC
Eddy-covariance method
RF
Radiative forcing
CBL
Convective boundary layer
ITCZ
Inter Tropical Convergence Zone
Table 0.1: Acronyms and Abbreviations in the order of reference.
2
Chapter 1
Introduction
The Earth system is in a thermodynamic state that is far from the equilibrium (Kleidon, 2010). Maintenance of thermodynamic disequilibrium can be attributed to several
factors. First and foremost is the Earth’s atmosphere, which traps longwave radiation,
and concurrently acts as the fluid medium in global circulation. Together with spatially
and temporarily varying insolation, the atmospheric characteristics enable a perpetual
maintenance of the horizontal and vertical temperature gradients within the system, a
clear indicator of disequilibrium. The atmosphere is by itself not unique to our planet,
as it is found in all the “gaseous” planets of the Solar system as well as on Titan and
Mars (Lorenz et al., 2001). Some other drivers of thermodynamic disequilibrium such as
primary production and atmospheric chemical reactions can be partially or exclusively
attributed to Earth’s unique property, which is that it is a living planet (Lovelock et al.,
1975). Its thermodynamic disequilibrium is further enhanced by physical and chemical
interactions between the biosphere and geosphere (Kleidon, 2012). All the aforementioned
processes contribute to the fact that Earth cannot be treated as a simple isolated thermodynamic system. In the presented study we will limit the analysis to the role of radiation
and turbulence in the thermodynamic disequilibrium. Thermodynamic disequilibrium
maintains a permanent temperature gradient between the surface and the atmosphere,
which is driving the vertical heat transport. Then, a part of the heat in converted to potential energy, which is the source of the power for the atmospheric and oceanic motions
(Ozawa et al., 2003). The process is quantified with two physical laws, namely, the first
and the second laws of thermodynamics better known as the energy conservation law and
the entropy law. The change in entropy of a system is defined as a summation of the
supplied heat divided by the temperature of the system (Eq. 1.1), (Ozawa et al., 2003).
dQ̇
(1.1)
T
Entropy production in the Earth System can be attributed to the radiative heating of
the surface and the atmosphere and to the mixing of heat due to the vertical transport. Besides the entropy production within the system, the entropy is also perpetually exchanged
dṠsystem =
3
across the system’s boundaries. The degree of heat exchange and entropy production is
crucial for the determination of the strength of the circulation in the thermodynamic
system. Paltridge (1979) suggested that the mean state of the present climate can be explained with maximum rate of entropy production due to horizontal heat transport in the
atmosphere and oceans (Ozawa et al., 2003). The maximum entropy production principle
(MEP) refers to the upper theoretical limit in the thermodynamic system. The application of the principle implicitly postulates that the system seeks to maximize its entropy.
Since then, numerous researchers have been exploring the validity of the MEP hypothesis.
Amongst the most promising fields of application is the turbulent heat transport (Kleidon
et al., 2006) and the large scale horizontal transport (Grassl, 1981). Application of the
thermodynamic limits represents a different physical approach in Earth systems modeling, when compared with the conventional climate and weather prediction models. When
solving for local and global energy balance closure, conventional models apply the law of
energy conservation, whereas the entropy in the system is not explicitly used, either as a
state variable or as a physical constraint (IFS Model cycle - Cy40r1, 2013). In that way,
disregarding the entropy in the conventional model frameworks increases the degree of
underdetermination in the system. These models resolve the issue with an introduction
of prescribed empirical parameters, which relate the exchange quantities, e.g. heat or
momentum, to the gradients of the driving variables. In the systems where the MEP hypothesis was found to be true, application of the limit eliminates the need for prescribing
the parameters. The thermodynamic limits therefore offer an opportunity to improve the
physical consistency and quality of the turbulent transport projections in the models.
The main topic of the thesis is the modelling of turbulent heat transport in a smallscale surface-atmosphere column within the diurnal scales. The working hypothesis of the
framework is that the turbulent transport at any given time of the day adheres to the
maximum power generation principle (MPP), which is a particular example of the MEP
principle. We provide a brief summary the selected research on the MEP hypothesis.
We then introduce the proposed MPP framework with three model illustrations, each
consecutive model adding an additional degree of complexity to the system. In the first
model illustration we briefly introduce the notion of the Earth system as a thermodynamic
machinery consisting of the atmospheric and surface heat reservoirs. Using the first and
second laws of thermodynamics, we explain how the disequilibrium state is maintained
and prove that as a consequence the Earth System can be treated as a heat engine. In the
second model, where the atmosphere and surface are assumed to be in a thermodynamic
steady state, we extend the framework with an atmospheric engine performing at the MPP
limit, which is driven by vertical heat transport. In the following subsection we outline
the final model framework with the energetic boundary layer extension, thus laying the
foundations for the radiative and convective heat engine model, a conceptual alternative
to the conventional climate models. Finally, we present the particular objectives of the
thesis research.
4
1.1
Applications of the Thermodynamic Limits in
Research
A wide spectrum of conducted research has shown that certain thermodynamic processes of the Earth system can be adequately modeled by employing the MEP or MPP
principle. Only a handful of articles shall be mentioned to provide a brief overview of the
research on the topic.
Most of the studies tested the adequacy of the thermodynamic limits for large scale
systems. With a simple zonally averaged energy balance climate model, Grassl (1981)
tested the hypothesis that large-scale meridional motions act in accordance with the
maximum entropy production. The MEP principle was used as a constraint to solve the
system of the four unknowns - meridional fluxes, surface heat fluxes, cloud cover and
temperature. The hypothesis showed satisfactory agreement between the model outputs
and observations with respect to the temperature, meridional heat fluxes and cloud cover.
Thermodynamic limits were also tested in more complex models. Kleidon et al. (2003)
conducted sensitivity simulations of an atmospheric general circulation model subjected
to the MEP constraint, by adjusting model resolution and friction to represent turbulence in the boundary layer. The authors emphasize the importance of using sufficiently
high model resolution and relevant boundary layer drag parametrizations to adequately
simulate the equator-poleward heat transport. The MEP limit also suggested that the
general circulation models (GCMs) tend to underestimate the atmospheric heat transport
and therefore overestimate the equator-pole temperature gradient (Kleidon et al., 2003).
It is remarkable how the application of the MEP principle enables even relatively simple
zonally averaged energy balance models to reproduce certain properties of much more
complex GCMs.
Application of the MEP principle in the zonally averaged energy-balance models was
also proven as a helpful approach on other planetary objects in the Solar system. It was
demonstrated by Lorenz et al. (2001) that meridional heat diffusion on Mars and Titan
is far better constrained by the use of the MEP, in comparison to a more complex but
commonly used parametrization in Earth system models, which is a function of surface
pressure, heat capacity and angular rotation. Furthermore, the MEP based approach also
results in a fairly good prediction of the annually averaged temperatures for the tropics
and polar regions of the planets. In Kleidon, Renner and Porada (2014), the applicability
of the MPP concept to study the Earth’s surface energy balance and hydrological cycle was
demonstrated, arriving at reasonable first-order estimates of the annually averaged surface
heat fluxes for a large number of observation sites, only using the solar radiation and water
availability measurements. Further on, the modeled evaporation fluxes and runoff across
major river basins matched reasonably well with the observations. The research indicates
the MPP principle can also be applied within smaller-scale systems (e.g., catchments/river
basin scale), if all the major energy flows and the annual hydrological cycle within the
boundaries of such systems can be inferred. Application of the MPP principle makes
5
it possible to model first-order estimates of the surface fluxes, using only a handful of
necessary measured variables at the site, namely the net solar radiation at the surface
and precipitation. This observation puts the MPP in a perspective, where it could become
a promising prognostic tool for the turbulent heat flux projections, especially in the regions
where observation data are scarce, since average solar radiation and even precipitation can
be determined by a use of remote sensing instruments or be even stochastically modelled
from climatological means.
The presented scope of research suggest that certain processes in the Earth systems
operate near their thermodynamic limits. In the systems where adherence to one of
the thermodynamic limits is clearly identified, this provides a valuable constraint to the
physical dynamics of the thermodynamic processes.
1.2
1.2.1
Thermodynamic Limits in Earth Systems Science
Thermodynamic Disequilibrium of the Earth System
Maintenance of thermodynamic disequilibrium can be illustrated with a simple two
reservoir model illustration, adapted from Kleidon (2010). The model consists of two heat
reservoirs, representing the surface and the atmosphere each at its respective temperature and heat capacity (Fig. 1.1). There is heat exchange with outer space, originating
from solar radiation Jin,s & Jin,a and emission of longwave radiation, either through the
atmospheric window Jout,s (surface reservoir) or atmospheric longwave emission Jout,a (atmospheric reservoir). Space can also be thought of as a boundary condition reservoir
at the constant temperature Tsun that is supplying energy to the Earth. Heat Jheat is
exchanged across the Earth’s reservoirs, from the warmer surface into the atmosphere. In
the absence of any work done by the system, the first law of thermodynamics yields an
expression for the change in the surface (Eq. 1.2) and atmospheric (Eq. 1.3) reservoir
temperatures Ts & Ta :
c·
dTs
= Jin,s − Jout,s − Jheat
dt
(1.2)
dTa
= Jin,a − Jout,a + Jheat
(1.3)
dt
The entropy production budget of the total system dSdttot consisting of both surface
dSs
a
and atmospheric reservoirs dS
is a product of three processes: entropy mixing with
dt
dt
space σmix,s & σmix,a , entropy mixing associated with heat transport between the surface
and atmosphere σheat and entropy exchange across the system’s boundaries N EE (Eq.
1.4). The entropy production components that are indicated as sigmas all share the same
functional formulation e.g. Eq. 1.7. The formulation represents a heat transfer from one
reservoir to another one, thus effectively energetically mixing the reservoirs.
c·
6
dSs dSa
dStot
=
+
= σmix,s + σmix,a + σheat − N EE
(1.4)
dt
dt
dt
Entropy mixing with space, i.e. with the sun, is calculated by as heat transfer Jin,s & Jin,a
between the respective reservoirs of the Earth and Sun, which is according to Kleidon
(2010):
σmix,s = Jin,s (
1
1
−
)
Ts Tsun
(1.5)
1
1
−
)
(1.6)
Ta Tsun
Entropy mixing by the heat flux Jheat ,which is exchanging heat between the surface
and atmospheric reservoir is written in a similar fashion (Eq. 1.7).
σmix,a = Jin,a (
1
1
− )
(1.7)
Ta Ts
So far we have characterized the entropy input into the system (Eqs. 1.5 and 1.6)
from the sun and entropy production within the system (Eq. 1.7). Finally, we must not
forget about the net entropy exchange (NEE) across the Earth system’s boundaries (Eq
1.8). The NEE is a difference between the outgoing entropy due to the Earth’s surface
and atmospheric longwave radiation and the incoming entropy from the solar radiation.
σheat = Jheat (
Jin,s + Jin,a
Jout,s Jout,a
+
)−(
)
(1.8)
Ts
Ta
Tsun
The formulation in Eq. 1.4 explains the underlying processes of the disequilibrium.
One can notice that disequilibrium is maintained due to energy exchange across the
Earth’s boundaries, the driving force of all the entropy production terms except for the
mixing heat term σheat . Were the Earth System isolated, the total entropy would reach
a maximum value, determined exclusively by the weakening heat mixing term. The heat
exchange between the surface and the atmosphere would deplete the temperature gradient up to a point where the system would have reached a thermal equilibrium. That is
not the case because the energy input from the Sun maintains the temperature gradient
in the vertical, thus facilitating the permanent export of entropy through the system’s
boundaries. Finally, the resulting heat flux and the prevailing temperature gradients are
a source of potential energy that drive the atmospheric and oceanic motions.
N EE = (
1.2.2
Maximum Power Generation of the Atmospheric Heat Engine
We can illustrate the ability of the atmosphere to preform work by describing it as a
heat engine. Atmospheric heat engine is an idealized system that illustrates the conversion
7
Figure 1.1: A simple two reservoir model. Schematic representation of the energy fluxes in the Earth
system that is in a state of thermodynamic disequilibrium. The system consists of the surface reservoir
with temperature TS and the atmospheric reservoir with temperature TA . The surface is warmed due to
absorbed solar radiation flux Jin,s and cooled by the outgoing longwave radiation (atmospheric window)
Jout,s , whereas the atmosphere is warmed by the solar radiation flux Jin,a and cooled by atmospheric
radiation Jout,a . The surface absorbs more energy from the Sun than atmosphere. This heating difference
results in a temperature difference between the reservoirs that drives the vertical heat transfer from a
warmer (TS ) to a colder (TA ) reservoir Jheat . Adapted from Kleidon et al., (2014).
of turbulent heat fluxes into convective motions. We follow the steps of the framework
presented in Fig. 1.1, extending it with such a heat engine, locked between the surface
and the atmosphere, which together with the heat engine constitute the bigger energetic
system (Fig. 1.2). The following part of the derivation only focuses on the smaller
heat engine system, which is why for the time being we omit the energy balance at
the surface and the atmosphere and the corresponding thermodynamic disequilibrium
equations. The system is assumed to be in a steady state, implying constant temperatures
of both reservoirs and no change in the internal energy and entropy of the system. This
assumption is only used at this point of the presentation in order to simplify the derivation,
whereas the steady state assumption is not applied in the final model framework. The
engine is powered by the heat flux, running from the surface Jheat . Like in any heat engine,
a part of the heat flux is used to extract power Pex . Also, one must not forget to include
the dissipative processes D in the atmosphere, which provide an additional heating source
for the engine. That is because the power, which manifests itself in convective motions,
is eventually dissipated within the heat engine and is therefore not extracted to be used
outside the heat engine system as is the case for a mechanical heat engine (Kleidon and
Renner, 2013). Furthermore, there is a justification why the dissipation term was fed
back to the the heat engine’s input. The friction is strongest close to the surface, which
is why the bulk of dissipation occurs in the lower part of the boundary layer, i.e. close to
the surface (Stull, p.155, 1988). Starting from the first law of thermodynamics, we obtain
8
the energy balance equation of the small heat engine system (Eq. 1.9). The heat inputs
Jheat and D are supplied to the engine from the surface, whereas Jout is the output at the
atmospheric reservoir, after a fraction of the surface heat was converted into power.
Jheat + D − Pex = Jout
(1.9)
At this point we apply the steady state assumption between the dissipation and power
generation Pex = D. Even though this assumption cannot be justified at diurnal time
scales, it is a simplification that has to be applied at this stage to obtain a determined
system of equations. Also, the power generation and dissipation terms are about an
order of magnitude smaller than the other terms in the energy balance equation, which is
why the simplification should not play a detrimental role in the model. The net entropy
exchange of the system consists of Jheat and D, added at a temperature Ts and Jout
removed at a temperature Ta (Eq. 1.10). Using the second law of thermodynamics
(Eq. 1.1), we know that maximum power is extracted from a smaller heat engine when
no irreversible processes take place within the system i.e. when the rate of entropy
production within the system and its exchange across the boundaries equals zero (Eq.
1.10) (Kleidon, 2010). Note that the maximum power extraction occurs when the entropy
production within the smaller system is at its theoretical minimum. However, this does
not contradict with the state of disequilibrium, which is preconditioned on a positive
entropy production of the entire energetic system. The disequilibrium within that system
is sustained and the entropy production remains positive (Eq. 1.6), as the MPP principle
adheres only to the smaller atmospheric heat engine system (Eq. 1.10 and Fig. 1.2).
Jheat + D Jout
dSsystem
=(
−
)=0
dt
Ts
Ta
(1.10)
Inferring Jout from Eq. 1.9 and inserting it into Eq. 1.10 we obtain the MPP limit (Eq.
1.11). The limit that adheres to the zero entropy production is commonly known as the
Carnot limit, even though the derived limit slightly differs from the original theoretical
Carnot limit (Kleidon, 2012). That is because in the presented model the power is not
extracted from the engine, but rather fed back into the system as heat in the form of
dissipation. It is evident from the Eq. 1.11 that power extraction is proportional to
the temperature gradient and the heat flux. These terms are in fact interdependent;
the temperature gradient is driving the convective heat transfer, which is at the same
time depleting the gradient. As much as both factors contribute to the maximum power
extraction they are jointly also its limiting factors.
Ts − Ta
(1.11)
Ta
We should keep in mind that the model’s simplicity implies the use of some essential
assumptions. Firstly, a steady state was assumed where dissipation equals the power
generation and the temperature gradient is in balance with the heat flux. Within diurnal
Pex = Jheat
9
time scales, particularly during the boundary layer evolution, such steady state is not an
adequate assumption (Kleidon et al. 2014). What one usually observes is a power-intense
buildup of boundary layer in the morning followed by a period of an approximate steadystate around the afternoon, before the dominating dissipation breaks down the boundary
layer at the end of the day. Also, a two reservoir model aggregates very distinct properties
of the atmosphere in only one atmospheric reservoir, thus failing to capture pronounced
diurnal cycles within the bottom part of the atmosphere, namely the boundary layer.
The described model also considers that the heat flux is only driven by the surface fluxes,
thus disregarding horizontal heat advection that occurs frequently with passing of the
fronts and other synoptic weather systems. This thesis is to a large extent an attempt
to prove these shortcomings can be overcome or be at least significantly reduced by an
introduction of an additional, energetic boundary layer reservoir. Nevertheless, even the
simplest thermodynamic limit frameworks have already proven its ability to accurately
model the dynamics of some essential planetary processes, as we have demonstrated in
the literature overview section.
Figure 1.2: A simple two reservoir model with a heat engine system. The scheme illustrates functioning
of the atmospheric heat engine that operates at the MPP limit. The heat engine is primarily driven by
the surface heat flux Jheat , a fraction of which is transferred into power P (represented with a violet
arrow) that drives turbulent convective motions. Convective motions are being dissipated at the surface
D (three violet arrows), thus providing an additional heat input in the engine. The remainder of the heat
flux that is not utilized in the conversion to power Jout is transferring heat to the atmospheric reservoir.
Adapted from Kleidon and Renner (2014).
10
1.3
Energetic Boundary Layer, a Heart of the Atmospheric Heat Engine
1.3.1
Energetic Boundary Layer Extension of the Two Reservoir
Model
An atmospheric heat engine is characterized by a rather peculiar fact concerning its
power source. Although entirely dependent on the incoming solar radiation, the engine
is not driven directly by that source. On the contrary, the atmospheric heat engine
is predominantly driven by the surface sensible heat flux which facilitates convective
exchange between the surface and the atmosphere, a consequence of the Earth’s surface
capacity to better absorb shortwave energy from the penetrating solar radiation than the
air. On a sunny day the surface heats up more efficiently than the atmosphere. This
results in a temperature gradient between the surface and the near surface atmospheric
layer which starts the turbulent heat transport in the vertical, thus warming the surface
atmospheric layer. With some delay the heated surface atmospheric layer transfers a
part of the received energy to the layer above. What we observed is a gradual heating
of the atmospheric column up to a point where the surface triggered heating completely
diminishes. We define that part of the atmosphere as the energetic boundary layer. Note
that the proposed definition of the boundary layer differs from the one applied in boundary
layer meteorology, where the boundary layer thickness exhibits a large temporal variability
on the diurnal scale. Alternatively, the energetic boundary layer (EBL) is defined as a
reservoir with a fixed upper boundary, which separates the atmospheric layer, which
internal energy is significantly influenced by the surface at any given time of the diurnal
cycle from the rest of the atmosphere (Kleidon and Renner, 2014). An educated guess,
based on the EBL observations at high latitude field stations informs us that a fixed
boundary should be set at approximately 400 hPa above surface pressure.
Differential heating and cooling in the vertical is an important property of the atmosphere that can not be represented by the simple model presented in the subsection
1.2.2. In that model the transfer of heat is assumed to take place uniformly from the
surface reservoir to the atmospheric one. In truth however, the atmosphere as a whole
is non-uniform in its thermal properties (Stull, 1988). Whereas the surface layer features
temperatures comparable to those of the surface, the top of the troposphere usually exhibits temperatures that are several tens of degrees lower. Therefore, it is reasonable to
partition the one uniform atmospheric reservoir into layers. We have partitioned the atmosphere into two reservoirs, the first one being the aforementioned EBL and the second
one top of the atmosphere (TOA). Hence, the aggregates the entire atmosphere above the
boundary layer into one reservoir. The EBL on the other hand is where differential diurnal heating takes place and thus also serves as a reservoir for the heat storage and release
Ha . The EBL is being heated during the day. Therefore, a part of the energy, supplied
by the surface, is stored in the EBL. At the night time however, the EBL gradually cools
11
Figure 1.3: An illustration of the EBL, an atmospheric reservoir positioned between the surface and
TOA. The EBL is a “small” thermodynamic system that performs as a medium of the atmospheric heat
engine, driven by the surface turbulent fluxes H & λE and the absorbed longwave radiation ( 34 τ −1)RL,atm .
Apart from facilitating the convective motions in the system, as described in Fig. 1.2, the EBL also stores
and releases heat. The heat storage of the system Ha is portrayed with a “measuring jug”, thus indicating
the energetic state of the system.
down and as a result, releases heat into the TOA. The EBL is also heated by longwave
absorption. Containing a high density of water vapor, the EBL plays an important role
in the greenhouse effect.
The EBL extension thus provides an additional reservoir to the framework. Due to
its heat containing properties, a new reservoir also affects the entropy production in the
system. Further implications and a complete derivation of the extended framework are
too extensive to be explained in the introduction and will therefore be presented in the
following chapters, namely. For now let us just keep in mind that by altering the entropy
production, the boundary layer extension also influences the power limit of the system,
which affects the partitioning of the turbulent fluxes. At first sight, an introduction of
the EBL heat storage looks only as a slight modification in the system. However, it turns
us to be an essential one, as it allows us to capture convective and radiative dynamics of
the system on diurnal time scales.
1.4
Objectives and Aims of Thesis Research
Numerous research studies, which have been previously discussed, indicate that application of the thermodynamic limits clearly has a potential in the Earth Systems modelling.
Yet, its scope and a particular use within more complex systems is still largely unclear
and therefore remains to be resolved. That determined a starting point which set me on a
path of trial and error to devise an alternative framework. From the start the research was
aimed at modelling land-atmosphere dynamics for small-scale ecological systems, centered
12
around two key aspects. The first main theme of study was the temporal dynamics of the
surface turbulent fluxes, which drive the land-atmosphere exchange, associated with vertical convection, while the second topic to be investigated was the use of the MPP concept
in the boundary layer growth dynamics. Early stages of research have shown that modelling of the boundary layer dynamics would require an additional degree of complexity
in the model associated with inevitable modelling of vertical temperature profiles which
was beyond my capabilities within a relatively narrow time frame. Consequently, the
research emphasis has moved towards the use of the MPP concepts as a tool to analyze
diurnal variability in surface energy fluxes. Nevertheless, as a consequence of a closed
energy balance in the atmospheric column we can besides the turbulent fluxes also get an
additional insight into the other energy fluxes in the column.
With their model, Kleidon, Renner and Porada (2014) have proven that the application
of the MPP provides reasonable first-order estimates for annually averaged surface heat
fluxes. A major goal of this research is to transform this simple model into a non-steady
state framework that will be able to project reasonable first-order estimates on much
shorter diurnal time scales. The centerpiece of the framework will be an atmospheric
column energy equilibrium model, where the EBL is treated as a convective and radiative
atmospheric heat engine with a temporally varying heat storage that performs at the MPP
limit. A step-wise modular design approach is proposed, where the model’s framework
is complemented with new sources and drivers of the land-atmosphere exchange, thus
improving physical consistency and precision of the model. The model will consist of the
following modules: a soil moisture module for calculating water availability, a soil heat
flux module and vegetation module for estimating stomatal conductance, a key regulator
of the latent heat transpiration (Jarvis, 1976).
The model’s performance will be evaluated by a comparison of the projected turbulent
heat fluxes with the observations from a high latitude field station in Finland, Hyytiälä.
After the quality assessment of the model projections, a potential further application of
the proposed framework to study surface energy exchange at other sites and also a possible
implementation of the MPP principle in the more complex general circulation models will
be discussed.
Special attention in our research will be given to an examination of the EBL heat
storage term Ha , a crucial part for the diurnal extension of the atmospheric heat engine.
The EBL heat storage is a fundamentally new concept in the land-atmosphere interaction
studies that was in a rudimentary form proposed by Kleidon and Renner, (2014). For
that reason, it is still largely unknown whether our perception of the storage term and
its temporal variability corresponds to the actual diurnal energy dynamics in the EBL.
Therefore, one of the aims is to validate the conceptual construction of the EBL heat
storage by comparing the modelled diurnal evolution of the storage term with an analytical calculation from radiosonde measurements. The model will also be used for studying
the effects of changing climatological conditions, especially a projected greenhouse enhancement on the partitioning of the surface fluxes, vertical transport and as a control
experiment, mean surface temperature. Conclusions from sensitivity analysis will be com-
13
pared with trends from the IPCC climate projections. A detailed physical explication of
the framework, including description of the selected modules and conducted experiments,
will be provided in the thesis report.
1.4.1
Research goals
1. To construct a physically consistent modelling framework and test whether the
atmosphere performs as a thermodynamic heat engine, subjected to the maximum
power limit.
2. To devise such an alternative approach to model surface turbulent heat fluxes, which
would require only a limited number of measured variables.
3. To study the effects of a changing climate on the partitioning of the surface turbulent
heat fluxes, vertical transport and mean temperature, using sensitivity analysis to
an enhanced greenhouse effect.
4. To explore a potential applicability of the maximum power generation limit in the
more complex models.
1.4.2
Hypotheses
1. Turbulent heat transport between the surface and the atmosphere can be interpreted
as the driving force of the atmospheric heat engine that is operating at its theoretical
maximum power generation limit.
2. Application of the maximum power generation principle in the atmospheric heat
engine leads to reasonable first order projections of the diurnal surface turbulent
heat fluxes, using only a limited number of generally accessible input variables.
3. Energetic boundary layer heat storage Ha captures the diurnal variations of energy
in the atmospheric layer that is energetically influenced by the surface.
14
Chapter 2
Modelling Framework
Aim of the Introduction chapter was to provide a brief explanation of the key thermodynamic processes behind the atmospheric heat engine. The outline was complemented
with an indication about the framework’s potential applications which will be assessed in
this thesis. The goal of this chapter is to further present the most salient features of the
model framework. We will start with a description of the column energy balance model,
constructed for the purposes of our research. Further on, the design of a soil heat flux,
a soil moisture and vegetation modules, which are coupled to the energy balance framework will be briefly presented. Key aspects of the extensive framework will be holistically
captured in a comprehensive model graphics scheme.
2.1
Equations of the Radiative and Convective Atmospheric Heat Engine with the MPP Limit
For the purpose of a holistic understanding and application of the framework, we now
make a step beyond the previously presented conceptual reasoning by quantifying the
flows of energy and entropy in the system. We start with energy balance equations at the
surface, energetic boundary layer and top of the atmosphere. When separating the energy
components, we follow the global radiation budget scheme by Trenberth et al., (2009). A
set of energy balance equations is presented with incoming components on the left-hand
side of equations and outgoing on the right-hand side. Afterwards, the MPP limit of the
extended framework is inferred from energy balance of the heat engine and the entropy
law (Kleidon and Renner, (2013) and Renner (2014)). Throughout derivation, it might
be helpful to keep the comprehensive model scheme in mind (Fig. 2.2).
2.1.1
Surface Energy Balance
3
(1 − ϕ)Rs + τ RL,atm = RL,surf + H + λE + Qg
4
(2.1)
15
Total absorbed solar radiation at the surface and in the atmosphere Rs is partitioned
into the part that is absorbed by the atmosphere ϕRs and therefore not available at the
surface, and the component which is absorbed at the surface (1 − ϕ)Rs (Eq. 2.1). An
additional energy input is provided by the atmosphere, which radiates in the longwave
spectrum as a gray body RL,atm = σTa4 (Nakajima et al., 1992). In the process of atmospheric downward radiation transfer the longwave radiation component is increased by
the radiative properties of the greenhouse gases (Inamdar et al., 1998). The downgoing
longwave radiation is commonly known as the greenhouse effect and is in our model parameterized as 43 τ RL,atm . This parametrization accounts for the greenhouse enhancement
of the incoming longwave radiation by applying a spectrally uniform Gray atmosphere
model with Eddington’s approximation, where we assign a multiplier that is proportional
to the optical depth τ of all the greenhouse gases (Nakajima et al., 1992). The details
of the Gray model are not essential in order to understand the column energy balance
model. The essential feature is that the optical depth in the radiative transfer relates to
the atmospheric ability to absorb and emit longwave radiation. Specific dependance of the
optical depth on greenhouse gases concentrations and other variables is beyond the scope
of this thesis study. The incoming fluxes are balanced by the outgoing longwave radiation
RL,surf and surface turbulent heat fluxes, namely the sensible and latent components H
and λE. Additionally, a small energy fraction is either being stored or released from the
ground by the soil heat flux Qg .
2.1.2
Energetic Boundary Layer Energy Balance
It was explained in the introduction that, besides functioning as a medium for extracting power, the EBL also stores and releases heat. For that reason the energetic boundary
layer heat storage term Ha was introduced. The storage term is a difference between the
heat flux input (λE + H + D + ( 34 τ − 1) · RL,atm ) in the heat engine and the sum of power
generation P and the outgoing fluxes Jout from the engine (Eq. 2.2).
3
λE + H + ( τ − 1) · RL,atm + D − P − Jout = Ha
(2.2)
4
It was also explained that the engine is powered by the surface turbulent heat fluxes
and the absorbed longwave radiation. The EBL is the most dense part of the atmosphere
with respect to water vapor, a highly absorbent greenhouse gas in the longwave spectrum, which is why the EBL is partially also heated and cooled by a recurrent absorption
and emission of longwave radiation (Held and Soden, 2000). The amount of longwave
radiation that is “trapped” in the boundary layer is not easy to determine since it is temporally highly variable as it depends on numerous factors, such as the water content and
its vertical distribution in the atmosphere, inclination of the sun and cloud cover amongst
others (Inamdar et al., 1998). In an elementary model such as ours, these factors are
far too complex to be explicitly included. For that reason, the absorbed longwave radiation term in the EBL was resolved with a simple approach. Knowing that the longwave
16
radiation emitted at the TOA equals RL,atm and that the incoming longwave radiation
at the surface equals 43 τ RL,atm , implies the difference in the terms comes from the the
greenhouse-enhancing property of water vapor. Water vapor is by far the most important
greenhouse gas in the atmosphere (Held and Soden, 2000) with most of the water vapor
content situated in the EBL part of the atmosphere. For that reason, the greenhouse
enhancement of the downgoing longwave radiation ( 34 τ − 1)RL,atm was attributed to the
water vapor in the EBL (Fig. 2.1). The source of the longwave radiative heating in the
boundary layer comes from the surface outgoing longwave radiation that is partially absorbed in EBL (Fig 2.1). Accordingly, we assume that the amount of the outgoing surface
longwave radiation, absorbed in the boundary layer, equals the greenhouse enhancement.
The proposed solution, depicted in Fig. 2.1, therefore follows a typical text-book illustration, portraying the greenhouse effect as a radiation recycling mechanism (Trenberth et
al., 2009).
When inserting the surface and TOA energy balance equations, presented in the following subsection (Eqs. 2.1., 2.4 and 2.7) in Eq. 2.2, the EBL heat storage term can be
written as a function of external forcing:
Ha = (1 − ϕ)Rs − Qg − (1 − ϕ)Rs
(2.3)
The latter expression enables a clearer interpretation of the heat storage term. According to Eq. 2.3, the EBL stores the remaining energy from the surface solar radiation
(1 − ϕ)Rs that is not absorbed by the soil Qg or released in outer space (1 − ϕ)Rs (Eqs.
2.3 and 2.7). Such formulation intuitively makes sense, as it presents the storage term in
a form of a residual reservoir, which implicitly stems from our definition of the energetic
boundary layer.
2.1.3
Top of the Atmosphere Energy Balance
The TOA reservoir receives energy from the surface RL,surf . Due to the greenhouse
effect, a part of the surface outgoing longwave is absorbed in EBL, thus decreasing the
absorbed longwave radiation in the TOA reservoir. The other source of energy for the
reservoir comes from the outgoing convective heat engine flux Jout . Additionally, the
atmosphere is heated up by a proportion of the total incoming solar radiation ϕRs . The
atmosphere releases received energy in the form of gray body radiation, which is according
to the Stefan-Boltzmann law a function of the reservoir’s temperature RL,atm = σTA4
(Inamdar et al., 1998). Radiation is emitted back to the surface as well as into space,
the downgoing component being enhanced by the greenhouse effect. The total outgoing
radiation from the TOA reservoir therefore equals (1 + 1 · 34 τ )RL,atm . A complete energy
balance equation for the TOA reservoir is shown in Eq. 2.4.
3
3
RL,surf − ( τ − 1) · RL,atm + Jout + ϕRs = (1 + τ )RL,atm
4
4
(2.4)
17
Figure 2.1: The scheme illustrates the EBL and TOA reservoirs that jointly constitute the atmosphere
and the longwave fluxes in the system. The outgoing longwave radiation at the surface RL,surf is partly
absorbed in the EBL and partly in the TOA. The TOA radiates to outer space RL,atm and back to
the surface 34 τ RL,atm . The downgoing longwave radiation from the TOA is enhanced due to radiative
properties of the greenhouse gases τ . (Adapted from Trenberth et al., 2009).
By averaging Eq. 2.4 and deducing the averaged expression from the same equation,
we obtain an additional constraint to the atmospheric energy balance (Eq. 2.5). The
derivation steps are a pure technical routine of applying daily averaging rules and are as
such hereby omitted.
RL,atm = RL,atm + ϕ(Rs − Rs )
(2.5)
A closed energy balance in the atmospheric column postulates that the diurnal average
of incoming solar radiation equals the daily average of outgoing longwave radiation at the
top of the atmosphere, therefore:
RL,atm = Rs
(2.6)
After inserting Eq. 2.6 into Eq. 2.5, we can see that the temporal atmospheric
longwave term equals the average solar radiation at the surface plus a variable proportion
of the total incoming solar radiation:
RL,atm = (1 − ϕ)Rs + ϕRs
(2.7)
Finally, we must address the simplifications, which were made in the derivation of
Eqs. 2.1 - 2.7, and the consequent limitations of the framework. We assume the whole
atmospheric column (EBL and TOA) as completely opaque, which implies that there is
no longwave penetration of the atmosphere. This simplification, which effectively neglects
the existence of the atmospheric window flux (Trenberth et al., 2009), was not intended
in the initial framework design. Unfortunately, adding the atmospheric window has made
18
the system underdetermined as we could not come up with an additional equation. We
thus complied with the opaque atmosphere assumption, which implications will be thoroughly addressed in the results and discussion chapters. Horizontal heat advection in the
atmospheric column is also neglected. An addition of that term would require a temporal
determination of the flux at the systems’s boundaries, requiring a multitude of measurements around the modelling observation site. For average meteorological conditions, the
horizontal advection is not the dominating term in the column energy balance. Grimmond and Oke (1995) supported this claim in an evaluation of the urban energy balances,
where horizontal advection plays a more important role due to the anthropogenic heat
flux. They conclude that horizontal advection is difficult to evaluate, but can probably be
neglected. We have also excluded the effects of clouds, which not only vary the partitioning of solar radiation absorption at the surface and atmosphere ϕ, but also significantly
alters the greenhouse effect (Trenberth et al., 2009).
2.1.4
Power of the Atmospheric Heat Engine
So far a set of energy balance equations was presented. Now we complement the set
by assuming that the atmosphere works as a heat engine performing at the MPP limit.
This proposition distinguishes the proposed framework from conventional energy balance
models, which do not (explicitly) apply the entropy law in the system closure. We start
by evaluating the entropy production in the heat engine. Heat transfer is analogous to
the system, described in section 1.2.2 (Eq. 1.10). With an introduction of the EBL this
intermediate reservoir is not just transferring heat from the surface to the atmosphere,
but is also exchanging heat as it is being warmed up or cooled down. For this reason, a
b
is introduced (Eq 2.8).
EBL entropy production term dS
dt
Jin Jout
dSb
−
=
(2.8)
TS
TA
dt
A general idealization of the boundary layer and atmosphere is that it consists of two
gases, i.e. dry air and water vapor, commonly assumed to adhere to an ideal gas law. The
entropy state of any ideal gas can according to Hauf and Höller, (1987) be written as:
S = cp ln(
p
T
) − R · ln
+ Sref
Tref
pref
(2.9)
where R is the gas constant, cp specific heat at constant pressure and the ref index an
arbitrary reference state of the system. Mathematically speaking the reference state is a
constant of integration which drops out in our analysis, since we are not really interested
in the absolute amount of entropy in the system, but rather in the entropy changes of the
system. The entropy can be expressed more compactly using the definition of a potential
R
temperature for moist air θi = Ti ( pp0i ) cp (Hauf and Höller, 1987). Inserting the latter
expression in Eq. 2.9, we get:
19
S = cp ln
θ
θref
+ Sref
(2.10)
The entropy state of the EBL is thus directly related to the potential temperature
for moist air (Eq. 2.10). As we know from boundary layer meteorology, the dynamic
boundary layer tends to have a non-static vertical profile of potential temperature, which is
also true for the EBL (Stull, p.284, 1988). A simple reservoir approach does not explicitly
allow such a degree of complexity. Therefore, a bulk potential temperature of the EBL
is introduced < θb >, a variable which represents an average potential temperature and
consequently also the energy and entropic state of the reservoir. With an introduction
of the bulk temperature, the EBL entropy production term can now be written as a
derivative of Eq. 2.10:
dSb
ρcp h · d < θb >
=
(2.11)
dt
< θb > dt
where h represents a static EBL height. The numerator in the equations actually
stands for the heat storage or release within the EBL. Besides being an indicator of the
entropy production in the system, the average temperature in the column< θb > is also a
representative value of internal energy in the EBL calculated as ρcp h· < θb >. Heating of
the boundary layer is provided by a sum of the surface sensible heat flux and greenhouse
longwave absorption ( 34 τ − 1)RL,atm , whereas the EBL is cooled down by the outgoing
heat flux (Eq. 2.12).
3
d < θb >
= H + ( τ − 1)RL,atm − Jout
(2.12)
dt
4
In absence of moisture condensation, the surface latent heat flux only provides moisture
into the system and should therefore not be included in the equation. Exclusion of the
latent heat flux is the only difference between the internal energy of the EBL and EBL
heat storage Ha , where latent heat is also included. We have assumed no condensation
in the entropy production equation because the physics behind the process is just too
complex to model. On days when there is very little or no precipitation, the released
latent heat does not play a role in the entropy production of the engine making this a
valid assumption.
We once more apply the steady state assumption between power generation and surface
friction dissipation P = D. The right hand site of Eq. 2.12 thus equals Ha − λE (Eq.
2.2). We insert the latter term in Eq. (2.12) to finally obtain an expression for the EBL
entropy production:
ρcp h ·
Ha − λE
dSb
=
(2.13)
dt
< θb >
We continue with an explanation of the left hand side of Eq. 2.8, which related the
entropy production in the heat engine with the heat transfer from the surface to the TOA.
20
The ingoing heat flux Jin is a sum of the greenhouse longwave absorption, surface sensible
heat flux and dissipation by friction D, which equals power generation P . The expression
for the outgoing heat flux Jout is inferred from Eq. 2.2. Thus, we obtain a detailed entropy
production equation:
H + P + ( 43 τ − 1)RL,atm H + λE + ( 34 τ − 1)RL,atm − Ha
Ha − λE
−
=
TS
TA
< θb >
(2.14)
In order to avoid an underdetermined system of equations, we assume that the bulk
temperature of the boundary layer approximately equals surface temperature < θb >≈ Ts ,
a valid assumption in a convective boundary layer model (Stull, p.442, 1988). Validity
of the latter assumption was addressed in the evaluation of the framework using field
station data, obtaining an average difference in the temperatures < θb > −TS = 3.4◦ C.
This justifies our approximation as we only underestimates the bulk temperature on the
order of 1 %. After a rearrangement of the terms (Eq. 2.14), where we put all the terms
except for P on the right-hand side of the Eq. 2.14, we get a simple expression for the
atmospheric power generation P , with a fairly straightforward interpretation (Eq. 2.15).
H + λE − Ha + ( 34 τ − 1)RL,atm
· (TS − TA )
(2.15)
TA
Power generation is primarily driven by the ingoing heat fluxes from the surface and
the absorbed longwave radiation from the surface. The EBL heat storage Ha acts as a
buffer that takes away a part of available heat during the daytime when Ha is positive,
thus decreasing the power generation and releases it into the system in the night when
the term is negative, thus sustaining the power generation of the system. The power
generation is also proportional to the temperature difference between the surface and
atmosphere, consistently with the Carnot heat engine model, in which the temperature
difference determines the power generation efficiency (Schmiedl and Seifert, 2008).
P =
2.1.5
Derivation of the MPP Limit
Having obtained the power expression (Eq. 2.15), we can proceed with the derivation
of the maximum power limit. The limit of the framework, including soil and vegetation
modules can only be solved numerically because of a non-linearity in the latent heat
flux parametrization (Eq. 2.38). For that reason, an analytical solution of the limit is
presented, using a representative conceptual control model. In the control model the
parametrization of the latent heat flux is linear, which makes it possible to analytically
solve the system of equations, thus enabling us a clearer interpretation of the framework’s
results. Sensible and latent heat fluxes in the control model are initially aggregated in
one term called the surface turbulent heat flux: J = H + λE. Hence, the MPP limit takes
a slightly different form (Eq. 2.16).
21
J − Ha + ( 34 τ − 1)RL,atm
P =
· (TS − TA )
(2.16)
TA
We have obtained an expression for the power generation that is dependent on the energy
fluxes in the EBL and the surface-atmosphere temperature difference. In the following
step we will look for the maximum of the expression, which is an ultimate constrain in
our system. The formulation presented in Eq. 2.16 does not allow the maximization yet,
because the functional form on the right side consists of two interdependent processes,
namely the temperature difference TS − TA and the turbulent heat fluxes. Fortunately
we can rewrite the temperature difference as a function of external forcings to the system
(Rs & Qg ) by a use of the surface energy balance equation (Eq. 2.1). In order to do so, we
linearize the net radiation at the surface by transforming it to a form, where radiation is
proportional to the temperature difference Rnet = RL,surf − 34 τ RL,atm ≈ (TS −TA ). Firstly,
the Stefan-Boltzmann law is used to calculate the longwave radiation components (Eqs.
2.17 and 2.18). The surface outgoing longwave component RL,surf is linearized, using
the Taylor expansion around the average diurnal surface temperature T0 (Eq. 2.18). A
radiative exchange parameter kr that equals the value of a first derivative of the RL,surf
at the average surface temperature is introduced to linearize the net radiative exchange
(Eqs. 2.19). In the following step, we rewrite the net radiation at the surface (Eq. 2.20).
As we can see, the linearization around the average temperature does not bring us to the
desired functional form yet. For that reason, we introduce the net radiation constant RL,0
(Kleidon, 2015) (Eq. 2.21), which finally allows us to write the net surface radiation as a
function of the temperature difference (Eq. 2.22).
RL,atm = σTa4
(2.17)
4
4
3
RL,surf = σTs ≈ σT0 + 4T0 (TS − T0 )
(2.18)
3
kr = 4σT0
(2.19)
3
3
(2.20)
RL,surf − τ RL,atm = σT04 + kr · (TS − T0 ) − τ σTA4
4
4
3
RL,0 = σT04 − 4σT03 (T0 − TA ) − τ σTA4
(2.21)
4
3
Rnet = RL,surf − τ RL,atm = RL,0 + kr · (TS − TA )
(2.22)
4
A linearized net longwave radiation balance at the surface enables us to factor out
the temperature difference in the surface energy balance, after inserting the expression
forRnet Eq. 2.22 in Eq. 2.1:
(1 − ϕ)Rs − Qg − J − RL,0
(2.23)
kr
By inserting the later equation into Eq. 2.16 the power generation can be rewritten
as a function of the surface energy balance (Eq. 2.24).
(TS − TA ) =
22
(J − Ha + ( 43 τ − 1)RL,atm ) · ((1 − ϕ)Rs − Qg − J − RL,0 )
P =
(2.24)
kr · TA
A basic calculus rule states that for a function to have a maximum, the first derivative
of the function must equal zero, whereas the second derivative must be negative. We
solve the expression for the first derivative over J, as we are interested in the maximum
power generation with respect to the surface turbulence fluxes, knowing that other variables are independent of J (Kleidon and Renner, 2013). The first derivative, given by Eq
2.25, is then used to obtain the optimum surface turbulent fluxes, corresponding to the
maximum power generation of the atmospheric heat engine (Eq. 2.26). In the case of the
complete model, the turbulent fluxes are partitioned into the latent and sensible component H & λE. In the model, the power generation is numerically derived with respect to
the vertical exchange velocity w, which regulates the convection interdependently with
the temperature difference (Eqs. 2.28 and 2.29).
−2J + (1 − ϕ)Rs − Qg − RL,0 + Ha − ( 43 τ − 1)RL,atm
dP
=
=0
dJ
kr · TA
(1 − ϕ)Rs − Qg − RL,0 + Ha − ( 43 τ − 1)RL,atm
Jopt =
2
(2.25)
(2.26)
Evidently, the optimum surface turbulent flux is powered by a half of available surface
solar radiation, half of boundary layer heat storage and soil heat flux and a half of the
longwave radiation terms −RL,0 − ( 34 τ − 1)RL,atm . A brief back of the envelope calculation
of the radiation terms shows that their contribution to the optimum turbulent flux is
positive under the expected values of the TA . With Eqs.: 2.1, 2.2, 2.4, 2.7 and 2.26 we have
obtained a closed set of equations that determine the stete of the atmospheric heat engine
at its maximum power limit. More importantly, the framework now makes it possible to
model the crucial processes and variables of the surface-atmosphere energy balance and
heat transfer (J, P, Ha , TS , Jout ), requiring only a limited set of external forcing data:
the soil heat flux and the absorbed solar radiation at the surface (Qg & Rs ) and some
model parameters. The simplicity of the this energy equilibrium framework in reflected in
the fact that the derived expressions determine the state of the atmospheric heat engine
and the dynamics of its energetic processes independently of the previous states of the
system. In order to calculate the energetic process in the system, we do not need to
solve the temporal differential equations. Instead, the temporal evolution of the state is
determined by the current state of the forcing variables and a slow transient response
of the storage and longwave radiation terms. Of course the simplicity of the framework
implies certain limitations, which will be thoroughly examined in the discussion chapter.
23
2.2
Surface Turbulent Heat Fluxes Parametrization
Surface turbulent heat fluxes in global circulation models and numerical models for
weather prediction are usually parametrized using bulk transfer relations, based on the
Monin-Obukhov similarity theory (Eq. 2.27).
H = cp ρ
1
(θS − θ) λE0 = λρ
1
(qS − q)
(2.27)
raH
raV
The theory postulates that both sensible and latent heat fluxes are driven by vertical
gradients of potential temperature and specific humidity near surface (Eq. 2.27) (Garratt,
p.39, 1992). On the other hand, the magnitude of the fluxes is limited by the aerodynamic
resistance ra , which is used to capture the properties of atmospheric stability and surface
roughness that inhibit or facilitate the fluxes (Garratt, p.55, 1992). Experimental theory
suggests that raH = raV , when surface sources of heat and vapor are coincident, which
is true for the bare surface (Garratt, p.54, 1992). Such parametrization is not directly
applicable to our model with its course vertical resolution of only three reservoirs. Therefore, uniformity of the sensible and latent heat fluxes between the surface and the EBL is
assumed. Moreover, the driving force of the fluxes is a difference in ambient temperatures
of the surface TS and top of the energetic boundary layer TT BL . In that way, we implicitly assume a linear temperature profile in the atmosphere. A more adequate alternative
would be to model vertical potential temperature profile and vertically varying heat fluxes.
Yet, the proposed solution is unfeasible to be applied in the three reservoir framework,
where vertical temperature profile can only be determined with the three reservoir temperatures. After linearizing specific humidity and, additionally, assuming saturated air
at the top of the boundary layer reservoir (Kleidon et al., 2014), we obtain the following
parametrization for the sensible and latent heat fluxes:
H = ρcp w · (TS − TT BL )
λE0 = λρw · fw · (qS − qT BL ) ≈ cp (ρw) ·
fw s
· (TS − TT BL )
γ
(2.28)
(2.29)
where w is a vertical exchange coefficient, fw is a water availability parameter, which
expresses the role of soil moisture availability in bare soil evaporation with a value in
the range: 0-1, and s saturation vapor pressure curve (Kleidon and Renner, 2013). This
parametrization yields adequate partitioning at times when large convective motions take
place, i.e., when a convective boundary layer is observed (Garratt, p.58, 1992). One should
keep in mind though that in principle the surface turbulent heat fluxes are not determined
by a difference between the surface and top of the boundary layer temperatures, but rather
by temperature differences near surface. Therefore, the modelled fluxes are expected to be
particularly inaccurate on shorter time scales and cannot be used in certain developments
of the boundary layer, such as temperature inversion near surface. Parameterizations
24
used in Eqs. 2.28 and 2.29 are thus a simplification, which averages out the temperature
gradients within boundary layer, inherent to the application of the simple reservoir models.
2.3
Soil and Vegetation Modules
Land-atmosphere exchange processes are largely dependent on the characteristics of
the surface. For that purpose, we have designed the soil and vegetation modules, which
are coupled to the atmospheric heat engine model. The soil module is used to model the
soil heat flux, used in the surface energy balance equation (Eq. 2.1). It is also playing a
role of a moisture reservoir, which provides vegetation or bare soil with water, available
for evapotranspiration. The vegetation module also exhibits a twofold functionality. First
of all, vegetation canopy acts as an intermediate water reservoir for evaporation on its
own. Secondly, it utilizes a simple ecosystem stomatal resistance model (Garratt, p.235,
1992), (IFS Model cycle - Cy40r1, 2013) that mimics the vegetation’s property of regulating moisture release into the atmosphere. In this section each module is only briefly
summarized. For a more detailed description of the modules and their analysis, we refer
to Appendix A.
2.3.1
Soil Heat Flux
The soil is assumed to be isotropically homogeneous in its conductive properties but
not in soil temperature Tsoil . That is why the soil is partitioned in 4 layers of some characteristic layer depth. A layer depth of 6 cm is proposed because temperature variations
at the depths higher than 25 cm are negligible on diurnal time scales (Garratt, p.117,
1992). The role of radiation and convection in the soil can be neglected (De Vries, 1975).
Fourier’s law for heat conduction characterizes the heat flux between two layers, where
ks is thermal conductivity (Garratt, p.117, 1992):
∂Tsoil
(2.30)
∂z
As one can infer from from Eq. 2.30, the soil heat flux Qg depends on the soil temperature gradient and conductive properties of the soil, reflected by its thermal conductivity.
The latter is a function of soil heat capacity, soil density and soil diffusivity (Farouki,
1986) according to (Eq. 2.31). Soil density ρs , specific soil heat capacity cs and thermal
diffusivity κs are besides being characteristic for individual soil types also functions of the
moisture content in the soil θi (Farouki, 1986). Surface temperature and moisture storage
Wi can therefore be considered as the only forcing terms of the soil heat flux.
Qg = −ks
ks = ρs · cs · κs
(2.31)
Temperature change in an individual soil layer is determined by the difference between
the incoming and outgoing soil heat flux. An implicit Euler method is applied for solving
the four layer system with differential equations (Eq. 2.32).
25
∂Tsoil
∂Qg
= ρs · cs ·
∂z
∂t
2.3.2
(2.32)
Soil Moisture
Soil moisture storage is a vital component of the water cycle that is necessary to
determine the thermally conductive properties of a soil. Moreover, the moisture storage
is also indirectly related to the potential evaporation at the surface, which is our model
parameterized with the water availability parameter fw . A modified implementation of
a bucket model by Manabe (1969) was used to calculate the moisture storage and water
availability. As the name speaks for itself, the bucket model quantifies moisture in the
soil in terms of the water level in the bucket. Maximum water level is specified by the
respective soil’s saturation value and depth of the soil profile. The problem of a single
bucket model is that weaker precipitation events usually only slightly alters the relative
saturation in the bucket whereas in reality the top of the soil gets considerably saturated
(Garratt, p.231, 1992). Consequently, a single bucket model response to precipitation or
drainage is very slow. Due to the long residence times of the single buckets these models
do not adequately deal with moisture changes on diurnal scales. Therefore, the soil was
differentiated into three buckets, interconnected by the groundwater flow. The depth of
the buckets was determined by the thickness of the respective soil horizon/layer and its
moisture saturation value: Wmax = dz · θsat .
2.3.2.1
Water Balance Equations in the Soil
Moisture content W is a measure of the total water column that resides in each soil
profile. It is related to the degree of saturation, a variable that tells us what proportion of
the soil is being occupied by water (De Vries, 1975). The degree of saturation is defined
as:
θi =
Wi
· θsat,i
Wmax,i
(2.33)
Here Wmax,i characterizes the maximum water capacity of the layer i. When more
water is added to the soil that would result in an enhanced runoff, which would remove
the additional moisture over time. θsat,i approximately equals the fractional proportion
of pores in layer i, which have a potential to be filled up with water (IFS Model cycle Cy40r1, 2013). Moisture contents of the three buckets are determined with a system of
interconnected continuity equations (IFS Model cycle - Cy40r1, 2013). A change in the
moisture storage for the surface bucket (bucket number one) is calculated as a sum of four
processes: throughfall, which stands for the precipitation that penetrates the canopy T h
and can also be called as vegetation runoff, root extraction rooti · E(t) and groundwater
flow Fg1 (Eq. 2.34).
26
W1 (t + ∆t) = W1 (t) − [root1 · E(t) + Fg1 (W1 (t)) − T h(t + ∆t)] · ∆t
(2.34)
For buckets two and three, the flow is only determined by a sum of groundwater flows
and root extraction. Root extraction describes moisture extraction by the vegetation in an
individual layer. In ecological models the term is better known as the sap flow (Garratt,
p.240, 1992).
Wi (t + ∆t) = Wi (t) − [rooti · E(t) + Fgi−1 (Wi−1 (t)) − Fgi (Wi (t))] · ∆t
(2.35)
Throughfall is an input variable, whereas evapotranspiration is among the outputs of
the model from the previous time step E(t).
2.3.2.2
Water Availability
The parameter that relates bare soil evaporation at the surface with moisture storage
is water availability fw (Kleidon et al., 2014). Water availability, with its value ranging
from 0 to 1, depends on the specific humidity gradient between the surface and top of
EBL qsat (TS ) − qsat (TT BL ). Specific humidity varies as a function of both temperature
and relative humidity at the surface for bare soil (Eq. 2.36). The expression was derived
by equating the model parameterization of the bare soil latent heat flux (Eq. 2.29) with
the “Monin-Obukhov parameterization” (Eq. 2.27).
fw =
2.3.3
rh · qsat (TS ) − qsat (TT BL )
qsat (TS ) − qsat (TT BL )
(2.36)
Vegetation
The evaporation flux presented in Eq. 2.23 is only valid in the cases of a bare soil
surface. However, the observations used in the model analysis have been conducted over
a densely vegetated surface. That calls for a different parameterization of the surface
evaporation flux the one that captures the distinct evapotranspirative properties of a
forest ecosystem. Water is an elementary physiological need for trees, which is why
vegetation does not evaporate as “indifferently” as a bare soil (Garratt, p. 132, 1992).
One can imagine plants as water retention reservoirs coupled with soil that tend to regulate
moisture transfer in a way that optimizes their gross primary production. As a result,
evapotranspiration from dry vegetation can be considerably smaller than the evaporation
from a bare soil, as it not only depends on water availability in the soil, but also on
other ambient conditions such as temperature, nutrients concentration and the incoming
short wave radiation amongst others (Farquhar et al., 1980). This calls for an additional
extension of the soil module. Still, it is desirable to keep the model as simple as possible,
which is why a simplistic isothermal canopy vegetation model was implemented. In this
27
model we account for a decrease in evaporation by introducing stomatal resistance rc , a
variable that represents opening of the stomata in the foliage, which facilitates moisture
exchange through the leaves (Garratt, p.236, 1992). Evapotranspiration of dry vegetated
area is therefore parametrized as (Garratt, p.236, 1992):
Ev = ρ
s
· (TS − TT BL )
γ(raV + rc )
(2.37)
A detailed description of the vegetation module is not essential to our research and
is therefore omitted from the report. A brief complementary explanation on the Jarvis
model can be found in Appendix A.
2.3.3.1
Extension of the Surface-Water Balance
Vegetation significantly alters the surface water balance, especially when it has a
thick canopy. The canopy’s foliage acts as an interception reservoir, which is why the
throughfall in vegetated areas usually does not equal precipitation (Ilvesniemi et al.,
2010). This requires an introduction of an additional vegetation canopy water reservoir
in the model (Eq. 2.34). The water balance in the canopy reservoir was adapted from
Garratt, (p.237, 1992) and is a function of precipitation P rec, which increases the water
content of the canopy, and the sinks: throughfall T h (vegetation canopy runoff ) and
evapotranspiration Ev , which deplete the reservoir. A maximum canopy water storage m
is deemed proportional to the canopy leaf area, characterized with the dimensionless leaf
area index LAI (Garratt, p.237, 1992):
m ≈ 0.2 · LAI
(2.38)
Wveg (t + ∆t) = Wveg (t) + [P rec(t + ∆t) − E(t) − T h(t)] · ∆t
(2.39)
Vegetation canopy runoff T h equals zero when Wveg < m, whereas it equals precipitation P rec when Wveg ≥ m. At times when the canopy reservoir is at its maximum
capacity, the vegetation evaporates as a bare soil E0 . When there is no moisture on foliage, canopy evaporates as a dry vegetation Ev . Finally, in the intermediate regime the
evapotranspiration is parameterized as a linear combination of both boundary condition
regimes. To sum up, we present evapotranspiration for all the wetness regimes of the
canopy (Eq. 2.40).
E=


E

h v



Wveg
m
E0
· E0 + (1 −
Wveg
)
m
; m = 0 dry canopy
· Ev ; 0 < Wveg < m partially dry canopy
i
(2.40)
; Wveg ≥ m wet canopy
So far we have only presented the parameterizations for evaporation over a bare soil
surface and evapotranspiration over a vegetation. Usually though, evapotranspiration
28
includes contributions from the vegetation as well as the soil. A convenient solution is
to present the total evapotranspirative flux as a linear combination of both partial terms
(Eq. 2.41). For example, if the modelling site is partially covered with vegetation (Cveg )
and partly with bare soil, the aggregate evapotranspiration is calculated as:
E = (1 − Cveg ) · E0 + Cveg · Ev
2.4
(2.41)
Schematic Representation of the Model
Figure 2.2: A complete scheme of the modelling framework illustrates all the key processes in the
system. It shows a three layer soil moisture module with a set of three interconnected buckets. On
top of the soil module lies the vegetation module which regulates stomatal resistance of the forest, thus
affecting the magnitude of the surface the latent heat fluxes. The scheme also illustrates the framework’s
energy fluxes, the EBL heat storage and energy balances of the reservoirs. The atmospheric heat engine
in the EBL is portrayed next to the heat storage bucket in the form of convective motions that are being
dissipated near the surface.
29
Chapter 3
Methodology
The primary objective of our study, the modelling of the vertical turbulent heat transport, was conducted with a model study of atmospheric thermodynamics. The modelling
was conducted in several stages. After finalizing the outline of the theoretical framework, which was presented in the previous chapter, a representative conceptual model of
the convective and radiative atmospheric heat engine was constructed. The conceptual
model, which incorporated the energy balance in the atmospheric column, but lacked the
dynamic coupling with soil and vegetation, was designed with the intention to qualitatively demonstrate the ability of the framework to simulate the observed diurnal dynamics
of the system. Upon a successful simulation of the diurnal dynamics with the conceptual
model, we added individually tested soil and vegetation modules, thus obtaining the operating model that interactively coupled the atmospheric heat engine with the surface.
The model experiments were performed in relation to the research questions presented in
the introduction. The bulk of the runs was performed with intention to validate the MPP
hypotheses and prove that the alternative approach has a practical potential. Finally, a
local sensitivity analysis of the model was performed to test the robustness and reliability
of the model and also to provide an additional climatological insight in the modelled system. Accordingly, the sensitivity analysis was designed in the form of the global warming
experiments.
3.1
Application of the Conceptual Model
The conceptual model was designed in order to test, whether the energy balance
framework with the MPP constraint, can reproduce the diurnal characteristics of the
energy fluxes and vertical exchange velocity in the atmospheric column. Additionally, the
surface temperature was simulated as a control test. Climatological site that was chosen
for the control model analysis was Barstow, California, a semi-arid location with plentiful
of insolation and scarce vegetation cover. With 11 mm of average precipitation in July,
the water availability is presumably very low (NCDC, 2015). A weak greenhouse effect
30
was prescribed with the greenhouse forcing parameter of τ = 1.5 due to a presumed low
vapor density in the atmosphere. A theoretical absorbed solar radiation signal Rs was
generated, assuming the albedo of the desert ∼ 0.35 and calculating sun movement at
the site for the 15th of July (Henderson-Sellers and Wilson, 1983). As a semi-arid area,
Barstow was assumed to be a relatively cloudless region. Therefore, we assumed that only
10% of the total absorbed solar radiation is absorbed in the atmosphere. An idealized
diurnal surface temperature signal which was based on the daily average, maximum and
the minimum temperature at the site for July was generated. That data was obtained
from the National Climatic Data Center archive (NCDC, 2015). The temperature signal
was used to calculate the soil heat flux at the surface Qg by using the Fourier equation
(Eq. 2.30). The water availability parameter fw was prescribed as a linear function
with a weakly negative slope (educated guess), ranging from 0.12 at sunrise to 0.08 at
sunset, the values that are characteristic for a very dry soil. A comparison of the modelled
temperature output to the generated temperature signal was used in combination with
the surface energy balance closure as an independent verification of the framework. The
energy fluxes in the column were only qualitatively assessed.
3.2
Observations
A set ob observation data has a twofold significance for the research. First of all,
a number of observation data is used in the form of input variables in the model and
secondly, some observation variables (H, E, TS ) serve as the comparison data for the
modelled output. Hence, the observations played a key role in all the modelling processes: initialization, model runs, verification and analysis. We used measurements from
Hyytiälä Forestry SMEAR-II station (Station for Measuring Ecosystem-Atmosphere Relations) with a temporal resolution of 30 minutes. The data is freely accessible online (AVAA
database, 2014). Additionally, we have used high-resolution radiosonde data which were
released from the station site as a part of the HUMPPA/COPEC field campaign (Ouwersloot et al., 2012). All the data was collected from 13/07/2010 to 10/08/2010. The
main reason for a relatively narrow time frame and a selection of only one observation
site is rarity of high-resolution radiosonde data. At least four daily radiosonde measurements were required for an approximate calculation of the EBL heat storage Ha , while the
standard synoptic stations only provide up to two radiosonde releases per day. The data
handling and application in the model is further described in the following subsections.
A complete list of the observation variables can be found in the tables of variables and
parameters (Appendix C).
31
3.2.1
Hyytiälä Forestry Field Station Data
3.2.1.1
Soil and Vegetation Module
Soil and Vegetation modules were individually tested and compared with the observations from the Hyytiälä field station. The following variables were used in the analysis: precipitation P rec, evapotranspiration E, soil moisture θ, soil heat flux Qg and the
temperature at 4.4m TS . The soil was classified as Haplic podzol, a common soil type
in Scandinavia. As a consequence, we applied the corresponding sandy loam empirical
parameters in the soil module (Ilvesniemi et al., 2010; Greve et al., 1998; Clapp and Hornberger, 1978). Estimates of evapotranspiration were obtained from the eddy-covariance
measurements. The latent and sensible turbulent fluxes were measured approximately
23 m above ground, which is about 10 m above the canopy (Ilvesniemi et al., 2010;
Description of the SMEAR II Station, 2014). In order to avoid numerical errors, the
precipitation was temporally down-scaled to the intervals of 30 seconds using uniform
distribution. Vertical root distribution root was adapted from the specified values for
evergreen needleleaf trees biome in the IFS Model cycle - Cy40r1 (2013). The seven-layer
soil profile from the observations was aggregated into three layers, corresponding with
the number of buckets in the model. Near surface temperature observations (T4.4m ) were
used to calculate the soil heat flux and temperature of each soil profile. Soil heat flux
observations were then used to calibrate the conductivity and specific heat capacity of the
soil. It was assumed that 70% of the evapotranspiration flux is provided by vegetation,
and the residual is accounted as bare-soil evaporation, which was inferred from satellite
images of the observation site (Description of the SMEAR II Station, 2014).
/
3.2.1.2
Solar Radiation Forcing
The proportion of solar radiation that is absorbed at the surface (1−ϕ)Rs was inferred
from the global radiation measurements taken 16 m above the surface. The observations
were corrected with an albedo of the pine forest (1 − α) = 0.93 (Betts and Ball, 1997).
An educated guess was made for the incoming solar radiation absorbed at top of the
atmosphere ϕRs . We started with an assumption that 20% of the total incident solar
radiation Rtot is absorbed by clouds and atmosphere, which is supported in the global
energy balance schemes (Lacis and Hansen, 1974; Trenberth et al., 2009; Ramanathan
and Vogelmann, 1997). It is also suggested that approximately 20% of the total incident
solar radiation is reflected at the top of the atmosphere (Ramanathan and Vogelmann,
1997) (Eq. 3.1). From these assumptions, which are demonstrated with Eqs. 3.1 and 3.2,
the parameter value ϕ = 0.26 was inferred.
(1 − ϕ)Rs = 0.6 · (1 − α)Rtot
(3.1)
ϕRs = 0.2Rtot
(3.2)
32
3.2.1.3
The Greenhouse Effect
The greenhouse enhancement in the model was represented by a constant optical
depth parameter τ . The parameter was calculated from an averaged observed incoming
longwave radiation at the surface L ↓ and the average outgoing longwave radiation Ra at
Hyytiälä (Eq. 3.3).
3
τ · RL,atm = L↓
(3.3)
4
3.2.1.4
Surface Turbulent Heat Fluxes
Surface turbulent heat fluxes H, λE were measured with the eddy-covariance instruments (EC) positioned slightly above the canopy. Both H and λE data sets were subjected
to filtering, which eliminated the corrupted data and the measurements collected at times
of temperature inversion (a noticeable negative sensible heat flux), because the model
cannot reproduce these situations.
3.2.2
Calculation of the Energetic Boundary Layer Heat Storage
The heat storage term Ha is primarily increased by both surface turbulent heat fluxes
(Eq. 2.2). In order to calculate the heat storage contributions from both turbulent heat
fluxes, we separated the term in two parts, the sensible HaSH and the latent heat storage
HaLH . Sensible heat storage was defined as a transient change of the EBL internal
energy QSH = ρdry · cp · T · z in the vertical over time (Eq. 3.4). By assuming a hydrostatic
∂p
=
equilibrium, which is a standard assumption in the atmospheric modelling science: ∂z
−ρdry · g (Stull, p. 84, 1988), we replaced the atmospheric height coordinates z with
pressure coordinates p (Eq. 3.4). The internal energy can be imagined as as an integral
of temperature with respect to height, multiplied by the specific heat capacity of the
energetic boundary layer (Fig. 3.1).
HaSH =
cp
∂T
cp
∂p
dQSH
= − (dp ·
) − (T · )
dt
g
∂t
g
∂t
(3.4)
The first part of Eq. 3.4 represents local heating in the atmosphere, while the second one stands for pressure tendency, which represents the contribution of the column’s
vertical expansion or compression, thus regulating quantity of the dry air in the column.
A brief back of the envelope analysis has shown that the second term tends to dominate the sensible heat storage when there is a slight temporal pressure change at the
surface (∼ 0.1 hP a). The pressure tendency is strongly dominated by synoptic changes,
which are independent of the local surface heating. Synoptic weather developments can
therefore introduce considerable noise in the HaSH . In order to solve this problem, we
have introduced a cutoff pressure thickness ∆p, hence imposing a fixed thickness of the
EBL column. By using the cutoff pressure we contain the atmospheric column within a
constant pressure difference which partly diminishes the importance of the second term
33
Figure 3.1: Illustration of the EBL sensible heat storage calculation. The figure portrays two vertical
temperature profiles at different times. The sensible heat storage HaSH is a difference in the internal
energy of the column over time, which can be calculated as an integral of the vertical temperature over
height, divided by the time difference. The difference in the integrals of two vertical temperature profiles
is illustrated in the figure as a coloured surface area. A positive contribution to the HaSH is indicated
with an orange colour, whereas the blue reflects a negative contribution to the term.
altogether (Eq. 3.5). This approach was also used for practical reasons. Because the
radiosondes do not cover the entire vertical extent of the atmosphere, as they eventually
break down, a reference point is needed for an execution of the comparative analysis of
the measured profiles. Equation 3.5 shows a numerical summation of the internal energy
difference over time up to the cutoff difference of the column ∆p:
HaSH =
 p −∆p

s
ˆ
cp 

Ṫ (p, t) · dp

dQSH
=
dt
g
ps
" n
X T (pi+1 , t + 1) + T (pi , t + 1)
cp
=
·
g · ∆t
i=1
2
· (pi − pi+1 ) −
n
X
T (pi+1 , t) + T (pi , t)
i=1
2
#
· (pi − pi+1 )
The EBL latent heat storage HaLH is a measure of changes in the moisture content in
the EBL (Eq. 3.6). By assuming a hydrostatic equilibrium and an ideal gas law for water
vapor e = ρv Rv T and dry air p = ρdry Rdry T (Stull, p. 81, 1988), Eq. 3.6 can be rewritten
to a more convenient form without the vertically varying densities of water vapor ρv and
dry air ρdry (Eq. 3.7). The term was then calculated as a vertical summation of the water
vapor pressure e multiplied by the heat capacity constants (Eq. 3.8).
(3.5)
34
QLH = λρv dz
QLH = −
HaLH
dQLH
=
=
dt
ˆps
(3.6)
λ · e · Rdry · dp
g · Rv · p
(3.7)
λ · Rdry · ė
dp
g · Rv · p
ps −∆p
λ · Rdry
·
=
g · Rv ∆t
" n
X ei+1 (t + 1) + ei (t + 1)
i=1
(pi+1 + pi )
· (pi − pi+1 ) −
n
X
ei+1 (t) + ei (t)
i=1
(pi+1 + pi )
#
· (pi − pi+1 )
(3.8)
The heat storage components were calculated from 146 irregular radiosonde measurements released at Hyytiälä. The cutoff thickness of 400 hPa was chosen as the column
thickness because of an insignificant diurnal variation of the internal energy and moisture content above that threshold. A maximum time step for an adequate calculation of
the total heat storage was determined as six hours. Vertical temperature profiles that
were further apart were excluded from the analysis. The calculated total heat storage
Ha = HaLH + HaSH represents an approximate value of heat, released or stored at the
midpoint of a given time interval.
3.2.3
Output Data Analysis
The modeled surface turbulent heat fluxes were paired with the corresponding observation values, thus forming the prime data set for the analysis. Scatter plots were
then used to portray the correlations between the observations and the outputs. By comparing Theil-Sew estimator analysis with a simple linear regression, it has been proved
that the data is not significantly skewed or homoscedastic which justifies a simple linear
regression analysis. The analyzed data set was unevenly distributed with more frequent
measurements around noon and less frequent ones in the morning, thus disqualifying the
possibility of ensemble averaging. For that reason the data was fitted diurnally. Variable
means were obtained from an integral over the diurnal distribution of each respective
variable (Eq. 3.9). The averaging was also used to compare the modelled surface energy
balance with the averaged measurements.
´ tsunset
f it (variable) · dt
variable = tsunrise
(3.9)
tsunset − tsunrise
The modeled EBL heat storage was compared to the calculated storage from the
radiosonde data and also to the analytically calculated term (Eq. 2.3), by applying the
35
same diurnal analysis as for the turbulent fluxes. Besides the aforementioned variables,
a less attentive analysis was conducted for the surface temperature TS , the soil heat flux
Qg and the soil moisture storage θ. Analysis of the latter two variables can be found in
the Appendix A.
3.3
Changing Climate Sensitivity Analysis
An explicit representation of the greenhouse effect in the model enables us to design
climate change experiments in a relatively straightforward manner. By varying the optical
depth τ we change the incoming longwave radiation component (greenhouse enhancement)
at the surface, which is a particular example of the one at a time sensitivity analysis, and
study the changes of relevant energetic variables. The climate sensitivity parameter Λ
characterizes a response of a given variable to the changes of the greenhouse effect (Eq.
3.10).
(3.10)
∆variable = Λ · ∆L↓
The reference point in the analysis is the current climate optical depth τ = 2.036 which
W
is calculated from an averaged incoming longwave radiation at the surface L↓ = 375.6 m
2
(Eq. 3.3). The parameter was varied within a selected range, corresponding to the nine
W
W
greenhouse enhancement scenarios:[L↓ − 2.8 m
2 , L↓ + 6.5 m2 ], which lower range roughly
corresponds to the pre-industrial radiative forcing (Myhre et al., 2013) while the upper
range corresponds to the 8.5 RCP (Representative Concentration Pathways) scenario by
the IPCC (Moss et al., 2010). This was done with deliberation to compare the sensitivity
analysis of our model with the selected climate models. After an execution of the first set
of experiments, we have realized that our definition of an enhanced greenhouse effect does
not directly relate to the IPCC’s definition of the radiative forcing (RF), which defines
the RF as: “the change in net downward radiative flux at the tropopause after allowing
for stratospheric temperatures to readjust to radiative equilibrium, while holding surface
and tropospheric temperatures and state variables such as water vapor and cloud cover
fixed at the unperturbed values” (Myhre et al., 2013). Therefore we omit from using the
expression radiative forcing in the analysis, but rather characterize the analysis as sensitivity with respect to an enhanced greenhouse radiation at the surface. The sensitivity
analysis was conducted on four characteristic variables: mean surface temperature TS ,
convective transport w∗ , sensible heat H and latent heat flux λE. Other terms of the
surface energy balance equation were also studied. Finally, we should be aware of the
limitations of our climate sensitivity analysis. First of all, as it was already explained, the
IPCC community executes the climate sensitivity analysis by using different definitions on
a change in radiative forcing. Secondly, the diurnal averages in our model were calculated
over the July-August measuring period, both being unconventional averaging periods as
the climate sensitivities are usually performed with respect to the annual or seasonal
averages. Moreover, our sensitivity analysis was performed on a specific land-based observation site. On contrary, a vast majority of climate sensitivity analyses is executed over
36
large-scale surfaces, therefore their values represent an aggregated sensitivity over various
types of land and even the oceans. Thus, we should keep in mind that our research is not
necessarily directly comparable with the conventional sensitivity analyses.
37
Chapter 4
Results
The modelling results are presented in three sections. We first present the results of the
conceptual model, which validate the applicability of the framework on the diurnal time
scales. We then evaluate a general performance of the complete framework by comparing
the diurnally averaged surface energy balance outputs with the corresponding analysis of
the observations from Hyytiälä. This is followed by an examination of the partitioning of
the surface turbulent fluxes. Diurnal averages are used to identify systematic differences
between the observations and the modelled results. Then the potential effects of the
horizontal heat advection, assumed as the most considerable systematic disturbance in
the model-observation comparison, are explored. In the following subsection we test the
concept of the EBL heat storage by comparing the modelled diurnal variation of the term
with calculated values from the radiosonde measurements and an analytical calculation
of the term from observed solar radiation and the modelled soil heat flux (Eq. 2.3).
Afterwards, the framework’s potential use for estimating the rates of vertical convective
transport is explored. Finally, we perform a sensitivity analysis of the model, which is
designed to validate the framework, but also with an intention to provide an insight into
the system’s response in temperature, partitioning of the surface turbulent fluxes and
changes in the rate of evaporation, all under the probable greenhouse enhancement in the
21st century. Sensitivity of the model will be compared with climate sensitivites of the
state of the art GCMs. A coherent interpretation of the results and their implications for
the validity of the hypotheses will be further developed in the discussion section.
4.1
Conceptual Model
The conceptual model was used to simulate a climatological July day in Barstow,
California (Fig. 4.1). Before we look into the results, we should mention an important
limitation of the model. Its design inherently limits the applicability of results to daytime
when the model correctly projects the heat transfer in the direction from the warmer
surface to the colder TOA reservoir. Negative values of the turbulent heat fluxes imply a
38
heat transfer from the cold to the warm reservoir, which contradicts with the second law
of thermodynamics (Eq. 1.1). Gray-shaded segments of the Fig. 4.1, which depict the
nonphysical results, are therefore omitted from the analysis. The implicit focus on the
diurnal time frame does not mean that the negative values of the surface heat fluxes are
negative. In fact, the negative values are plausible in the case of a vertical temperature
inversion. However, they are unphysical in our modelling framework, where due to a
course vertical resolution the temperature difference that drives the turbulent heat fluxes
is always positive.
A symmetry in the forcing functions as well as the outputs, plotted in the upper left
of Fig. 4.1 clearly demonstrate a high degree of idealization in the system. This is a
consequence of neglecting physical processes such as clouds, horizontal advection of heat
and moisture, condensation of moisture, explicit radiative warming and cooling in the
EBL and turbulent dissipation. In other words, solar radiation RS and the soil heat flux
Qg are the only energy inputs in the system. Consequently, the results resemble their
diurnal symmetry as well. The outputs must therefore be treated as a rough first order
approximation of the typical diurnal cycles in energy fluxes and other terms of the system.
The temperature control experiment, depicted in the upper right of the figure indicates
that the model reproduces the net absorbed energy at the surface reasonably well. The
modeled temperature amplitude is only slightly exaggerated. A greater inconsistency can
be observed in an overestimation of the morning warming and cooling at dawn. The
discrepancy can be largely attributed to two factors, 1) an inadequate representation of
the surface solar forcing (1 − ϕ)Rs , which is in reality a more sinusoidal-like function than
the dome shaped as depicted in Fig. 4.1, and 2) a disregard of the surface heat capacity,
which prolongs the time it takes to warm or cool the surface. The inferred energy fluxes in
the column capture the observed diurnal cycle reasonably well (Fig. 4.1, bottom left). Net
longwave radiation between surface and atmosphere Rl reflects the temperature differences
of the reservoirs. The surface turbulent heat fluxes are within the range expected for a
semi-arid location. Due to a relatively low water availability fw , the sensible heat flux
was calculated for a further evaluation of the
H is dominating. The Bowen ratio λE
H
framework, its values being in accordance with semi-arid surfaces (Garratt, p.36, 1994).
Optimum sensible and latent heat fluxes, which are inferred from the maximum power
constraint (Eq. 2.20) are strongly dependent on solar radiation whereas the other terms
in the equation contribute less energy. The modelled EBL heat storage Ha resembles
the characteristics that were illustrated in the introduction. During daytime the heat
storage is typically positive, indicating that the EBL is storing a significant amount of
energy. Around 6 pm, after solar heating is weakened, the EBL starts to release heat,
thus powering the heat engine in the absence of surface turbulent heat fluxes (Eq. 2.26).
Finally, let us look at the vertical exchange velocity w, the driving variable of the
surface turbulent fluxes (Eqs. 2.22 and 2.23). The vertical exchange velocity characterizes
the dependence of the vertical heat transport on the vertical temperature gradient. As a
consequence, the shape of the velocity plot reflects the diurnal evolution of the turbulent
fluxes (Fig. 4.1, bottom right). Vertical exchange velocity is a proportionality parameter
39
Figure 4.1: The conceptual model outputs. The four figures represent the diurnal-cycle of the conceptual
model outputs for a single climatological July day in Barstow, California. Model inputs are shown in the
dashed plot-lines, wheareas the outputs are shown in the solid plot-lines. The gray-shaded areas of the
figures indicate the non-physical night-time results. The upper left panel depicts the forcing functions in
the system: the absorbed radiation components Rs and the modelled soil heat flux Qg . The upper right
part of the figure shows the modelled diurnal surface temperature next to the generated climatological
temperature signal and the modelled atmospheric temperature. The bottom left panel shows the selected
energy fluxes in the atmospheric column, including both surface turbulent fluxes H & λE, the net radiative
exchange at the surface Rl and the EBL heat storage Ha . Finally, the bottom right panel vertical exchange
velocity coefficient w, a variable that is related to the intensity of the vertical convective transport.
for vertical transport of heat and moisture. However, this does not mean it can be directly
related to the convective velocity w∗ , a scaling parameter for vertical turbulent transport
in a convective boundary layer (Stull, p.188, 1988), even though both variables share
the appropriate units. Nonetheless, these parameters are closely related which is why
the modelled vertical exchange transport (Fig. 4.1, bottom right) closely resembles the
diurnal dynamics of convection (Eq. 4.3), (Stull, p.118, 1988).
40
4.2
Radiative and Convective Atmospheric Heat Engine Model
Whereas the conceptual model is only designed to provide a general insight in the
energy exchange processes, thus confirming a general physical consistency of the framework’s equations, the complete model aims to provide reasonable estimates of the energy
fluxes at the selected observation site at any given point of the day. Different objectives of
the models are also reflected in their complexity and the properties of the input data. The
conceptual model calculations are executed with the generated input data, which were
based on the monthly mean climatological values. The complete model on the other hand
projects the energy exchange on a basis of the conducted measurements in the system.
4.2.1
Surface Energy Balance
Analysis of the surface energy balance components is a crucial prerequisite in testing
the quality of the model. The evaluation was executed by comparing a diurnally averaged
energy balance of the model with the averaged observations from Hyytiälä (Fig. 4.2).
A notable overestimation in the model can be observed in the outgoing longwave
(+85.3 W/m2 ) and incoming longwave (+63.6 W/m2 ) components. Thus, the model
projects that the surface receives and radiates approximately 20% more than observed
in the measurements. This inaccuracy can be attributed to the opaque atmosphere assumption in the framework. Remember that we neglected the existence of an atmospheric
window, which reflects the role of a direct longwave radiation release to space. In Trenberth et al., (2009) a global average of the atmospheric window is estimated at 40 W/m2 ,
a value that is comparable to the overestimation in our model. A higher absolute increase in the outgoing longwave compared to the incoming longwave component results
in a decrease in the total turbulent fluxes. The sum of turbulent fluxes is therefore underestimated in the model (−17.7 W/m2 ), but the more pronounced difference lies in the
partitioning of the turbulent fluxes, with the model projecting a higher latent flux compared to the observations, whereas the observations indicate a higher sensible flux. A
more detailed comparison of the partitioned heat fluxes will be explained in the next subsection. Also as a consequence of an additional energy input at the surface, the projected
mean surface temperature is substantially overestimated (∆Ts = 14.8 K ).
41
Figure 4.2: Surface energy balance - measurements and the model. The graphics portray the diurnal
means of the surface energy balance fluxes and the mean surface temperature TS . The left panel scheme
portrays the means at the Hyytiälä measuring site, whereas the right panel figure depicts the means from
our model. The model clearly overestimates both longwave radiation components RL,surf & 34 τ RL,atm
and the latent heat flux λE, but on the other hand underestimates the surface sensible heat flux H.
4.2.2
Surface Turbulent Heat Fluxes
We present the results of the diurnally averaged fluxes and then evaluate correlations
between observations and the outputs. Correlation between the absorbed solar radiation
at the surface (1 − ϕ)Rs and the turbulent fluxes is also studied.
Figures 4.2b and 4.3b clearly show that the model reproduces the diurnal cycle. There
appears to be a time delay (phase shift) in the observations in regard to the model in
both figures. The shift is not uniform, as both sensible and latent fluxes peak after 11 am
local time. It can be inferred from the figures that the shift is a consequence of a quicker
buildup of the modelled surface turbulent fluxes in the morning, as well as their faster
decrease in the afternoon compared to the observations. Comparable standard deviations
of the monthly averages (box-plots) indicate that the model captures the extent of the
intermonthly flux variations reasonably well. The figures also show a noticeable difference
in the magnitudes between the observations and the model outputs. Projected latent heat
fluxes are higher, whereas the sensible fluxes are lower than in the measurements. That
aspect is more clearly demonstrated in the left panels of the figures (Figs. 4.3a and 4.4a),
where all 674 model results are shown jointly with the corresponding measured values
including a 1:1 line and the linear regression fit. Coefficient of determination for the both
2
2
components shows linear trends with noticeable scatter RλE
≈ 0.57 and RH
≈ 0.70. The
diurnal averages show a mean underestimation of projected surface sensible heat flux by
29%, whereas the surface latent heat flux comparison shows an average overestimation by
15%.
42
Figure 4.3: Sensible heat flux analysis. A comprehensive comparative analysis of the sensible heat
fluxes in the model and at the observation site. a) depicts all the 674 model runs, which are paired with
the corresponding observed values at the site (Hobs , Hmodelled ). The scatter plot and the linear regression
fit are portrayed next to a 1:1 (orange) line, to point out the difference between the observations and the
modelled values. b) shows an average diurnal cycle of the observed (red) and modelled (orange) sensible
heat flux. The box plots show standard deviations of the modelled and measured values at a given diurnal
time.
Figure 4.4: Latent heat flux model analysis. A comprehensive comparative analysis of the latent heat
fluxes in the model and at the observation site. a) shows all the 674 model runs, which are paired with the
corresponding observed values at the site (λEobs , λEmodelled ). The scatter plot and the linear regression
fit are portrayed next to a 1:1 (orange) line, to point out the difference between the observations and
the modelled values. b) depicts an average diurnal cycle of the observed (blue) and modelled (green)
latent heat flux. The box plots show standard deviations of the modelled and measured values at a
given diurnal time. An observed shift between the observation and the model plot suggests a premature
morning buildup and an afternoon breakdown of the temperature gradients in the model.
43
4.2.2.1
H and λE Dependence on RS
The derivation of the optimum surface turbulent heat flux indicates a strong correlation
with the surface solar radiation, since half of the radiation is used to directly power
the turbulent fluxes in the heat engine (Eq. 2.20). Figures 4.4a and 4.4b show the
diurnal evolution of the turbulent heat fluxes as a function of radiation, thus confirming
a strong interdependence with a correlation coefficient of r = 0.88 for the modelled fluxes
and r = 0.80 for the observations. Yet, a closer inspection of the figures points out
a clear distinction between the observed and modelled values including a hysteresis in
the diurnal cycle of the fluxes. The hysteresis is clearly visible in the measurements
(Fig. 4.5b), where we can observe morning fluxes occurring at higher solar radiation
values compared to the the afternoon values. That points that there needs to be stronger
radiation in the morning, in order to initialize the temperature and moisture gradients,
which drive the turbulent fluxes, than to sustain the gradients in the afternoon. The
model fails to reproduce the hysteresis as it actually makes a contrary projection with
an unpronounced reverse hysteresis i.e., there needs to be a slightly stronger radiation in
the afternoon to sustain the turbulent fluxes. This tendency was already observed in the
diurnally averaged turbulent heat fluxes where the morning fluxes are overestimated and
the afternoon fluxes underestimated compared to the observations (Figs. 4.3b and 4.4b).
The hysteresis in the observations can be explained with the transient response of the
observation site to the solar heating. Early in the morning the sun is relatively weak, so
it takes some time (about 1 hour, according to the Fig. 4.4b) for the surface, the canopy
and near surface layer of air to warm up sufficiently and initialize turbulent heat fluxes.
In contrast, in the late afternoon when the gradients are sustained relatively longer due
to the heat storage at the surface and the near-surface boundary layer. The framework
clearly does not reproduce this transient response adequately. The misrepresentation is
indicated with a higher correlation coefficient for the simulated surface flux and solar
radiation dependence compared to the observed relationship between surface fluxes and
radiation. The first and foremost candidate for the discrepancy is the heat capacity of the
vegetation canopy, because that factor is completely neglected in the framework. Heat
storage in high canopies, as is the case for Hyytiälä, may reach several tens of W/m2 ,
which approximately equals the difference in the morning latent heat values between the
model and the observations (Garratt, p.120, 1994). The overall effect of the storage
terms, which represent the dynamic responses in the system, therefore still seems to be
underestimated in the model framework. Moreover, a combination of modelled soil heat
flux Qg and the EBL heat storage Ha fails to reproduce the observed hysteresis. We
also should not dismiss other possible factors that contribute to the hysteresis and are
not incorporated in the model, such as: the apparent higher probability of clouds in the
afternoon (Fig. 4.6b), which are indicated by a reduction in the averaged Rs . The latter
could offer an explanation for the reverse hysteresis in the model output. Lower Rs in the
afternoon also implies a lower modelled heat storage Ha (Eq. 2.20). Consequently, the
heat storage contributes less energy to the turbulent heat fluxes in the afternoon than in
44
the morning (Figs. 4.6).
Figure 4.5: Turbulent heat fluxes dependence on solar radiation. Dependence of the surface turbulent
heat fluxes on the absorbed solar radiation at the surface. a) The reverse hysteresis of the modelled
turbulent heat fluxes suggests that an initialization of the fluxes requires less energy from the solar
radiation than its sustenance in the afternoon. This is opposite to the observations b) that in indicate a
stronger radiation is required in the morning, in order to build up the gradients than to sustain them in
the afternoon. The figures also suggest the interdependence of the turbulent fluxes and radiation in the
model is overestimated.
4.2.3
Energetic Boundary Layer Heat Storage
We conduct the analysis of the EBL heat storage with a two-fold comparison of the
term. Firstly, we compare the heat storage calculated from the radiosonde data with the
analytical storage that has been inferred from the modelled soil heat flux Qg and observed solar radiation Rs (Eq. 2.3). The comparison is then extended with the diurnally
averaged modelled heat storage. The averaged heat storage, calculated from the available
radiosonde data is presented together with the analytically calculated heat storage in a
whiskers plot (Fig. 4.6a). We observe that the mean values, which are indicated by circles, approximately fit the analytical curve, but the interquartile range is substantial and
most of the outliers lie beyond the vertical axis range (Fig. 4.6b). The interquantile range
of observations that can exceed 1000 W/m2 implies that a difference of about 100 W/m2
still means a reasonable match of the data sets. Naturally, changes in the boundary layer
heat storage cannot exceed the sum of the heat inputs H + λE + ( 34 τ − 1)RL,atm in the
system. Ranges of the third quartile in the whiskers plot (Fig. 4.6a) at 11 am and 5
pm therefore, together with most of the upper whisker ranges, appear to be non-physical.
However, the results are not really non-physical, they just clearly demonstrate the limitations of a local closure of the system, when there is in fact a significant contribution to the
energy changes in the column that is coming from the horizontally advected air masses
45
with different temperature and moisture properties. This discrepancy demonstrates a
troublesome nature of calculating the heat storage from the radiosonde data, which was
already indicated in the methodology chapter. Nonetheless, a relatively adequate match
in the mean values and the probability distribution of the interquartile ranges indicates
that there is a normal distribution of the non-local heating and pressure disturbances.
Comparison of the analytically calculated storage term to the averaged modelled storage
also shows a reasonable match (Fig. 4.7). Analogous to the modelled turbulent heat
fluxes, we observe a slightly quicker buildup of the storage in the morning, and a more
rapid demise of the term in the afternoon.
Figure 4.6: Energetic boundary layer heat storage analysis. a) Comparison of the EBL heat storage
calculations from the radiosonde data (whisker plot) and the forcing terms (Qg & (1 − ϕ) · Rs . The
whiskers plot shows a wide inter-quartile range of the radiosonde calculations with the outliers even out
of the figure’s range. b) The analytical EBL heat storage is plotted together with the forcing terms and
the mean values from the radiosonde heat storage (indicated by circles).
46
Figure 4.7: Average diurnal cylcle of the modelled EBL heat storage. Comparison of an average diurnal
cycle of the observed (violet) and modelled (orange) EBL heat storage. The box plots show standard
deviations of the modelled and measured values at a given diurnal time. A close agreement of the plots
confirms the physical consistency of the EBL heat storage implementation in the model.
4.2.4
Vertical Exchange Velocity
The diurnally averaged vertical exchange velocity roughly reproduces a typical diurnal
shape of the convective velocity scale w∗ in a convective boundary layer (Figure 4.8.),
(Stull, p. 119, 1988). The variables however differ in the magnitude. A typical magnitude
for the convective velocity on a very turbulent afternoon can be in the order of 1 m/s to
2 m/s (Stull, p.118, 1984), whereas the vertical exchange velocity in the model ranges
from 4 · 10−3 m/s to 5 · 10−3 m/s. By equating an averaged convective boundary layer
(CBL) parametrization with our parametrization of the sensible heat flux, we can relate
the vertical exchange velocity with the convective velocity scale (Eq. 2.28) (Stull, p.
118, 1988). This step facilitates an approximate modelling of the average rate of vertical
convection (Eq. 4.1). The value of the Von Karman constant is set at k ≈ 0.4 (Stull, p.181,
1988), and the average reference height in the dynamic boundary layer is put at one half of
the boundary layer thickness z = h2 . Finally, by inferring an average ambient temperature
difference in Hyytiälä (TS − TT BL ) = 30 K and guessing a potential temperature difference
in a convective boundary layer θS − θT BL = 3 K, we roughly approximate the proportion
−TT BL
of the temperature difference as TθSS −θ
≈ 10. This leads us to a first order conversion
T BL
from w to w∗ (Eq. 4.2). Using the conversion, we can see that the model suggests
an average convective velocity scale of 0.8 ms at noon, which is a reasonable first order
projection.
k · w∗ z · (1 − hz )2 ·
(θS −θT BL )
h
= w · (TS − TT BL )
(4.1)
47
.
w∗ = 200 · w
(4.2)
Figure 4.8: Vertical exchange velocity. A diurnal evolution of the modelled vertical exchange velocity
closely resembles that of the convective velocity scale w∗ , thus demonstrating that our simple framework
can be used to provide first-order estimates of the convective transport.
48
4.2.5
Surface and Atmospheric Temperatures
It was shown in the surface energy balance analysis that the model overestimates the
mean surface temperature T S (Fig. 4.2). The diurnal comparison of the modelled and
observed surface temperatures reveals that besides the magnitude, the diurnal amplitude
of the modelled temperature variation is also seriously exaggerated (Fig. 4.9a). The same
outcome applies to the modelled atmospheric temperature (Fig. 4.9b). Even though the
model was not designed with an intention to model temperature, this inconsistency points
out to a possible major shortcoming of the framework. The underlying reasons for the
inconsistency will be presented in the discussion section.
Figure 4.9: A mean diurnal cycle of the surface TS and atmospheric temperatures TA . a) Comparison of
the modelled diurnal surface temperature evolution with the observed temperature evolution at Hyytiälä.
We can see that the model drastically overestimates the amplitude of the cycle. b) The same conclusion
can be made for the atmospheric temperature of the TOA reservoir. The box plots in both figures
suggest an unnaturally high daily variability of the modelled temperatures (up to 35°C). In comparison,
the observed variability of the surface temperature ranges only up to 5°C.
49
4.3
Changing Climate Sensitivity Analysis
By performing the changing climate sensitivity analysis we explore the response of
the systemic variables to an altered radiative forcing at the surface. The design of the
executed climate experiments is simplistic, because we only change one systemic variable,
namely the optical depth τ that regulates the greenhouse effect. Such basic approach is
applied because the framework relies in a considerable extent on the real time observations
from the field station. Climate change effects on soil water availability, the soil heat flux,
relative humidity at the surface and other environmental factors, such as the physiological
properties of vegetation are non trivial and consequently difficult to model. As a detailed
systemic climate analysis was not one of our research goals, we did not design climate
scenarios for the aforementioned variables and systemic environmental factors, which were
determined with the observation data, but rather kept them constant. Still, in the process
of testing robustness of the model we conducted a one at a time sensitivity analysis
to study the response of the model to the changes in a selection of the environmental
parameters: ϕ, ρs , θsat , rmin , κs and Cveg . In that respect, the analysis suggested that
minor perturbations of the systemic environmental factors only result in a second-order
climate response of the system. On the other hand changes in the greenhouse enhancement
and surface temperature represent the first order responses to a changing climate. As we
shall see, a change in a greenhouse effect alone significantly alters the dynamics of the
system, first and foremost in the processes that depend on the surface temperature. The
sensitivity analysis is executed in the following steps. We firstly conduct a general analysis
of the surface energy balance under a changing climate. This is followed by a sensitivity
analysis of the mean surface temperature TS with respect to an enhanced greenhouse
effect. Furthermore, we present the sensitivity analyses of the surface turbulent fluxes
and convective exchange, both with respect to TS , in order to alow a comparison with
the climate sensitivities of the GCMs. EBL heat storage formulation in our model is not
sensitive to an enhanced global warming, as it only depends on solar radiation and the
soil heat flux (Eq. 2.3). Therefore the sensitivity analysis was not performed on that
particular variable.
4.3.1
Surface Energy Balance
Sensitivity of the surface energy budget to the greenhouse effect was tested using
a range of nine enhanced greenhouse effect scenarios, as mentioned in the methodology.
Here we only present the results of the lowest and the highest greenhouse forcing scenario,
because these two examples fully capture the trend of energy redistribution between the
fluxes at the surface. Figures 4.9a and 4.9b show the difference in the fluxes with respect
to the current climate (Fig. 4.2b). The first observation for both scenarios is that the sum
of the turbulent fluxes, as well as the net longwave radiation Rnet = RL,surf − 43 τ RL,atm
at the surface, is not affected by the changes in the greenhouse effect. Both longwave
components show a distinct increase in the ∆L↓ = 8.5 W/m2 scenario and a decrease in
50
Figure 4.10: The schemes illustrate the responses of the surface energy balance fluxes and the mean
surface temperature ∆TS to the changes in the greenhouse effect. The left panel of scheme presents
the responses in a decreased greenhouse forcing, roughly representing the pre-industrial state of the
system. The right panel presents the responses in a significantly warmer climate. The illustrations
suggest that an enhancement of the greenhouse effect leads to an increased longwave radiation components
RL,surf & RL,atm and the latent heat flux λE on one hand, and a decrease in the sensible heat flux H on
the other hand.
the ∆L↓ = −2.8 W/m2 scenario. A comparison of the depicted scenarios also shows that
global warming is refleced in a shift from the sensible heat flux H to the latent heat flux
λE. Trends of an increase in both longwave radiation components as well as in λE and
a decrease in H under enhanced greenhouse are consistent with the trends reported by
Boer (1993), Gutowski et al., (1991) and Andrews and Forster (2009), who independently
performed climate sensitivity analyses on the ensembles of the GCMs outputs. There is
however less agreement with the studies regarding absolute changes in the fluxes. The
referenced studies agree that an enhanced incoming radiation does not implicitly lead to
an equal enhancement of the outgoing radiation component, as suggested by the model.
The same holds for the turbulent fluxes. In fact, the studies suggest that for a doubling
in CO2 , the increase in the incoming longwave radiation exceeds the outgoing radiation
increase by ∆Rnet = [1 W/m2 °C, 1.4 W/m2 °C]. Thus provided excess heat at the surface
is mostly released with an enhanced latent heat flux (Andrews and Forster, 2009). The
sensible heat flux sensitivity is in comparison significantly smaller.
4.3.2
Mean Surface Temperature
We can infer from Fig. 4.10b that as a result of a greenhouse enhancement of 8.5 W the
surface is projected to get warmer by 1.36°C above the current mean surface temperature.
This is a significantly lower value than the mid range of the warming, projected by the
equilibrium climate sensitivity which is likely between 1.5◦ C to4.5◦ C for a doubling of CO2
concentrations, which represents the radiative forcing of 3.7 W/m2 (Collins et al., 2013;
Myhre et al., 1998). The underestimation is reflected in a lower climate sensitivity Λ of
the model Λ = 0.21°C/W m−2 , when compared with the mean value of the equilibrium
51
Figure 4.11: Sensitivity of the Mean Surface Temperature to a Changed Greenhouse Forcing. The
figure shows temperature responses of the model to a changed greenhouse forcing in relation to the current
states. A linear temperature response and a significantly lower climate sensitivity value Λ = 0.27◦ C/W
in comparison to the mean equlibirium climate sensitivity from GCMs Λ = 0.81◦ C/W demonstrates that
our simple framework disregards some of the crucial positive feedbacks in the climate system.
climate sensitivity range Λ = 0.81°C/W m−2 (Collins et al., 2013).
In the following step the sensitivity analysis of the greenhouse enhancement was extended up to a value corresponding to a 3.7◦ C warming, the mean projected temperature
increase by 2100 in the RCP 8.5 scenario (Collins et al., 2013). This enabled us to perform
the sensitivity analysis of the surface fluxes and the convective transport with respect to
a mean temperature change, which is a common approach when presenting climate sensitivities of the other variables.
4.3.3
Surface Turbulent Heat Fluxes
Changes in the surface turbulent fluxes in the global warming scenarios can be attributed to the projected alteration of the surface energy balance and higher temperatures
at the surface (Collins et al., 2013). The studies suggest a considerably higher sensitivity
of the latent heat flux compared to the sensible heat flux. For that reason the analysis will
mostly address the global warming effects on evaporation. The most important effect on
surface evaporation is the saturation water vapor pressure s (Eq. 2.23), which would on
average increase the evaporation at a rate of 6.5% ◦ C −1 (Boer, 1993). The saturation vapor pressure sensitivity is an upper limit of a relative increase in evaporation that can only
be achieved under a sufficient availability of moisture and energy for evaporation (Boer,
1993). Therefore, the projected changes in evaporation cannot be adequately made just
by looking at the temperature dependence of the saturation vapor pressure. The reported
range of sensitivities from climate models with ∆λE from 1.7 W/m2◦ C to 2.2 W/m2 °C
supports the existence of the processes that inhibit the potential increase of evaporative
52
Figure 4.12: Sensitivity of the surface turbulent heat fluxes to a mean surface temperature change. The
figure shows the surface sensible and latent heat responses to an increased mean surface temperature TS .
In absolute terms the sensitivity of the latent heat flux equals Λ = 3.16 W/°C, whereas the sensitivity of
the sensible heat flux equals Λ = −3.16 W/°C. It is also demonstrated that in relative terms, a decrease
in the sensible heat flux is larger than a relative increase in the latent heat flux, a consequence of the
smaller current value of the sensible heat flux compared to the latent heat flux.
rates (Boer, 1993; Gutowski et al., 1991; Andrews and Forster 2009). The modelled sensitivity of evaporation ∆λE = 3.16 W/◦ C agrees well with the reported sensitivities from
the climate models (Fig. 4.12). A more detailed analysis showed that ∼ 90% of an increase in the model can be contributed to a change in saturation vapor pressure, and the
remainder directly to an increased surface temperature. An increase in ∆λE in the model
is compensated by an equal decrease of the sensible heat flux ∆H = −3.16 W/◦ C. The
literature also projects a decrease in the sensible heat flux, however the projected decrease
∆H ranges from −0.2 W/m2 °C to −0.7 W/m2 °C and is consequently significantly smaller
than our model projections (Boer, 1993; Gutowski et al., 1991; Andrews and Forster
2009). In relative terms the evaporation is projected to increase by 2.8% °C −1 , whereas
the sensible heat flux is projected to decrease by −4.1% ◦ C −1 . This is comparable with the
study of Kleidon and Renner (2013), who performed a similar study with an even simpler
MPP framework, and obtained the relative sensitivity of evaporation of 2.2% ◦ C −1 .
53
4.3.4
Convective Transport
Mean convective transport, which is related to the vertical exchange velocity in our
model w is a very important characteristic of global climate. Our claim stems from
the fact that the deep convection in the tropics is the driving process of vertical moisture
transport in the Inter Tropical Convergence Zone (ITCZ), the upwelling part of the Hadley
cell, which decisively co-regulates the global climate system (Holton, p.378, 1973). The
climate sensitivity of convective transport is largely uncertain, as the convective transport
is not explicitly modelled in the comprehensive climate models (Betts, 1998; Held and
Soden, 2006). Still, there are studies that suggest a likely decrease in the convective
transport. Betts (1998) presents a simple reasoning for this trend that is applicable to
our model as well. The explanation is based on a parametrization for the latent heat flux
that is very similar to ours (Eqs. 2.29 and 4.3). We adapt his reasoning to the particular
design of our framework.
λE = λρw · (qS − qT BL )
(4.3)
The climate models project a relative climate sensitivity of the latent heat flux that is
smaller than the relative sensitivity of the saturation mixing ratios between the surface and
the top of the dynamic boundary layer qS − qT BL . This argument can be supported with
our analysis and literature review in the previous section, where we have explained that
the saturation vapor pressure climate sensitivity equals 6.5% ◦ C −1 . Climate sensitivity of
the mixing ratios is directly proportional to the saturation vapor pressure and therefore
also has a relative sensitivity of 6.5% ◦ C −1 (Boer, 1993). A smaller proportional change in
the surface latent heat flux (left hand side of Eq. 3.3) compared to the mixing ratios (the
right hand side) therefore requires a decrease in another control variable. The only variable
that can change because average density of the air remains approximately constant is the
vertical exchange velocity w. Deriving from a simple radiation-convective equilibrium
model, Betts and Ridgway (1989) therefore suggest a decrease in convective transport,
associated with a change in temperature on the order of −6.6% ◦ C −1 . Very similar values
are suggested by Held and Soden (2006) with −7.0% °C −1 and Kleidon and Renner (2013)
with −6.7% °C −1 . The relative sensitivity of the modeled convective velocity scale equals
−6.0% °C −1 and therefore comes close to the reported values.
54
Figure 4.13: Sensitivity of the convective velocity scale to a mean surface temperature change. The
figure shows an absolute and a relative response of the convective velocity w∗ to the mean surface
∗
temperature change TS . The modelled relative sensitivity w4w
= −6.0% ◦ C −1 is in agreement with
∗ ∆TS
the provided literature overview. The suggested relative sensitivity implies a significant reduction in the
mean convective transport under the temperature increases projected in the RCP climate scenarios by
the end of the century:1.0◦ C to 3.7◦ C, (Collins et al, 2013).
55
Chapter 5
Discussion
Even though the proposed framework is extremely simple, the results demonstrate it
can provide adequate estimates of the diurnal energy exchange in the land-atmosphere
interactions. However, the empirical analysis alone does not provide a sufficient reasoning
of the more fundamental question of how the system is regulated and why, if that is the
case, there is a tendency in the system to maximize its power generation. Hence, we
begin the discussion by providing a clarification of the processes that determine the MPP
state of the atmospheric heat engine. This is followed by a holistic evaluation of the
alternative framework in projecting the diurnal surface turbulent heat fluxes. We then
discuss the implications of the results on the EBL heat storage concept. The implications
section is concluded with an overview of the consequences from the changing climate
sensitivity analysis. The simplicity of the system naturally comes at a cost of limitations,
some of which are presented in the following section. After identifying the limitations, we
outline some possible improvements in the design for future work. Finally, we evaluate
the prospective use of the MPP limit in the more complex models.
5.1
5.1.1
Implications
The MPP Mechanism
In the first hypothesis we postulated that the surface turbulent heat fluxes drive the
atmospheric heat engine which can be assumed to operate according to the MPP limit.
The correlation of the modelled projections and the observations, presented in the results
section, strongly suggests this is the case but it cannot be interpreted as a sufficient proof
of the former hypothesis. After all, the fact that the system is transporting the same
amount of heat as a system in a MPP state does not prove that the system is necessarily
seeking such a state (Ozawa et al., 2003). For that reason, the MPP and related extremal
principles hypotheses have been dismissed by some as coincidental correlations (Ozawa
et al., 2003; Goody, 2007). With the intention to remove the shroud of mystery that has
up to the point surrounded the MPP concept in our research we present a qualitative
56
explanation adapted after Kleidon and Renner (2013) and Ozawa et al., (2003), which
provides reasoning of how is the system brought to a state of the maximum power generation. In this illustration, we propose the existence of a self-regulating MPP mechanism
in the atmospheric heat engine. The working hypothesis of the atmospheric heat engine
is that when turbulent fluxes transport heat from a warm TS to a cold reservoir TA a
part of that heat is converted into kinetic energy. The power generation, given by Eq.
2.15, is proportional to the sum of the surface turbulent fluxes J = H + λE and the
absorbed longwave radiation ( 34 τ − 1)RL,atm minus the EBL heat storage Ha , multiplied
by the temperature difference of the reservoirs (TS − TA ). The EBL heat storage Ha and
the longwave radiation ( 43 τ − 1)RL,atm can be treated as independent of the temperature
difference as well as of the amount of turbulent heat transport (Eqs. 2.3 and 2.17). Therefore, these terms can be treated as constants with respect to the intensity of turbulent
heat transport. Consequently, the power generation varies with respect to the turbulent
heat transport multiplied by the temperature difference (Eq. 5.1).
P ≈ J · (TS − TA ) + const · (TS − TA )
(5.1)
These are in fact two interactive properties of the system. Whereas the temperature
difference on one hand drives the turbulent heat transport (Eqs. 2.28 and 2.29), the
transported heat on the other hand regulates the temperature difference, thus implicitly
affecting the heat transport. The interdependence in the power equation (Eq. 5.1) thus
suggests a strong interaction of the surface and the atmosphere (Kleidon and Renner,
2013). The possible states of heat transport and associated power generation range from
the static case with no turbulent heat transport where J = 0 & (TS − TA ) = max to the
extreme mixing case where J = max & (TS − TA ) = min with the MPP somewhere in
between these two cases (Fig. 5.1). The main question is why would the system strive
towards the MPP state? Figure 5.1 illustrates the proposed self-regulating feedback mechanism that suggests an explanation for this tendency of the system. Imagine the system
in a non-maximum state A, shown on the left side of the figure. A small positive turbulent fluctuation +4J would increase the power generation in the system (Eq. 5.1). An
increase in the power generation leads to an increased convective transport, represented
by the vertical exchange coefficient w, which leads again to an increased turbulent heat
transport J because H & λE ∝ w. Therefore, the fluctuation tends to develop a positive
> 0, which continues to increase the convective transport
feedback mechanism where dP
dJ
and turbulent heat fluxes until the system reaches the MPP state (Fig. 5.2a). On the
other hand, if the system is in a non-maximum power generation state B, a positive turbulent fluctuation triggers a negative feedback which suppresses the fluctuations (Fig 5.2b)
(Ozawa et al., 2003). In contrast, a negative turbulent fluctuation −∆J triggers a positive feedback between the temperature difference and an enhanced power generation (Fig.
5.1), which again drives the system towards the MPP state. The proposed mechanism
explains the system’s tendency towards the MPP state, which is a single stable state of
the system and therefore provides justification for our first hypothesis. Still, we should be
57
aware that this is only a qualitative suggestion of a possible regulation mechanism in the
system that still needs further empirical and theoretical study to be properly validated
(Paltridge, 1979; Ozawa et al., 2003; Lorenz, 1960).
Figure 5.1: Schematic illustration of the proposed self-regulating MPP mechanism, showing power
generation as a function of surface turbulent heat fluxes J and temperature difference (TS − TA ). A
positive turbulent fluctuation at A and a negative fluctuation at B trigger a self-perpetuating increase in
power generation towards the stable MPP state. Adapted from Ozawa et al., (2003).
Figure 5.2: Feedbacks in the proposed self-regulating MPP mechanism. If an increase in one variable
leads to an increase of the other variable, this positive relation is denoted with an orange arrow and a +
sign, while the negative relation is illustrated with a blue arrow and a - sign. a) A positive feedback drives
the system from the state A (Fig. 5.1) towards the MPP state. b) A negative feedback suppresses the
positive turbulent fluctuations +∆J in the state B. On contrary, a negative fluctuation −∆J initializes
a positive feedback towards the MPP state.
58
5.1.2
Alternative Approach for the Modelling of the Surface
Turbulent Fluxes
A key defining characteristic of the proposed alternative approach is its simplicity. The
framework design is capable of modelling the energy dynamics of the atmospheric column
with the focus on the near-surface turbulent heat transport. In addition, the framework
is capable of modelling the diurnal dynamics of convective transport. What distinguishes
our approach from conventional models is the ability to describe the system without an
explicit quantification of the motion, described by the Navier-Stokes equations (Stull,
p.90, 1988) or alternatively, by the semi-empirical parametrizations of the the boundary
layer dynamics (Garratt, p.49, 1992). Furthermore, the state of the system is calculated
independently of the previous system’s states and only depends on the current state of the
forcing variables and the energetic system response captured by the EBL heat storage.
Finally, the most advantageous feature of our system is reflected in its ability to model
all of the aforementioned processes by using a very limited number of input variables
which can be easily obtainable with remote sensing instruments. The diurnal temporal
variation of the surface turbulent fluxes in the framework is a function of two energy
sources, firstly the radiative terms, which consist of the absorbed solar radiation at the
surface (1−ϕ)Rs and longwave radiation terms −RL,0 −( 34 τ −1)RL,atm , and secondly by the
heat storage terms Ha and Qg (soil heat flux actually represents the heat storage/release
of the soil) (Eq. 2.20). The turbulent heat fluxes are therefore partly regulated by
the the solar radiation terms and partly by the transient response of the surface and
the EBL represented by the storage terms. Comparative correlation analysis between
the surface turbulent heat fluxes and solar radiation (Fig. 4.5) suggests that the model
overestimates the role of solar radiation, even though the observations confirm a high
dependence of the fluxes on solar radiation. Nonetheless, the results are a clear indication
of an improvement over a more simplistic long-term average framework by Kleidon, Renner
and Porada, (2014), where the surface turbulent fluxes were entirely dependent on the
solar radiation Jopt = R2s , as a consequence of the heat storage extensions. Moreover, we
are confident that the overestimation of the solar radiation could be decisively eliminated
with an introduction of a canopy storage term, which is potentially a significant energy
contribution in a tall pine forest canopy at Hyytiälä (Garratt, p.120, 1994).
The model undoubtedly reproduces the observed diurnal cycle of the surface turbulent
fluxes reasonably well (Figs. 4.3b and 4.4b), but on the other hand performs less optimal
when it comes to the magnitudes of the fluxes (Figs. 4.3a and 4.3a). As it was explained in
the results chapter, the sensible heat flux is underestimated almost by a third, whereas the
latent heat flux is overestimated by 15%. There are three factors that may have affected
the performance of the model. First and the foremost is a coarse vertical resolution of the
two reservoir atmosphere, which implied the use of the parameterizations in which the
turbulent fluxes are proportional to the temperature difference between the surface and
the top of the EBL. We implicitly do not actually model the surface turbulent fluxes but
instead obtain the value of a uniform turbulent heat transport in the EBL. In other words,
59
as a consequence of a linearized temperature vertical temperature profile, we project the
vertically averaged turbulent heat fluxes in the EBL. If we take a look at the mean
characteristics of the turbulent heat fluxes within a convective mixed boundary layer
(Fig. 5.3), we clearly see that the vertically averaged sensible heat flux is considerably
smaller than its surface value whereas an averaged latent heat flux is higher than its surface
value. Therefore, the characteristics of the convective boundary layer (CBL) confirm our
explanation that the model outputs cannot be strictly regarded as the surface fluxes.
Another indicative reasoning for a discrepancy in the modelled and observed magnitudes comes from a previously conducted data analyses at the Hyytiälä field station
and other Scandinavian pine forest sites (Ilvesniemi et al., 2010; Suni et al., 2003). In
the article Water balance of a boreal Scots pine forest by Ilvesniemi et al. (2010), the
hydrological cycle of the site was thoroughly analyzed in the years 1998-2006. Among the
most interesting findings is an observation from a number of field studies in Scandinavia
that the energy balance could not be closed by the eddy covariance (EC) measurements,
possibly because the method underestimates evapotranspiration (Ilvesniemi et al., 2010).
The authors therefore suggested that the EC measurements underestimate the fluxes on
the order of ∼ 10% and probably up to 40% annually. The EC method at Hyytiälä has
been compared with the residual method where all the other water flows have been measured. The residual method infers higher annual fluxes up to 78%, though the method
lacks reliability (Ilvesniemi et al., 2010). On the basis of these conclusions, a negative
instrumentation bias should be assumed for the measured evapotranspiration values. The
authors suggest that the sensible heat flux is probably also subjected to a negative instrumentation bias but do arrive at an estimate. Due to significant standard deviations
the suggested instrumentation bias and the time-scale methodological differences in the
analysis by Ilvesniemi et al., (2010), the suggested corrections were not used to correct
our observation data set. Still, for the purpose of a qualitative analysis we can assume
that the observed surface turbulent heat fluxes should be evidently increased.
Finally, the third factor is a probable flaw in the framework design that is related to
the greenhouse radiation term in the EBL. After we had already conducted the model
runs and completed most of the model analysis, it was discovered that the proposed
concept of greenhouse absorption in the EBL is not physically consistent. However, this
inconsistency does not seriously affect the qualitative implications of the results. The flaw
will be described in more detail in the following section. At this point it should only be
noted that a correction of the faulty greenhouse term leads to a slightly altered expression
for the sum of optimum turbulent heat fluxes (Eq. 5.2).
(1 − ϕ)Rs − Qg − RL,0 + Ha
(5.2)
2
The new expression only differs from the previous one (Eq. 2.26) by effectively excluding the negative greenhouse absorption term, which reflects the assumed net longwave
absorption in the boundary layer. Elimination of the term results in a net increase of the
surface turbulent fluxes J. On average, the change implies an increase of the turbulent
Jopt =
60
Figure 5.3: Schematic illustration of the mean turbulent heat fluxes within an idealized CBL. Horizontal
axes are normalized with respect to the surface values of the heat fluxes, whereas the vertical axes are
normalized with respect to the top of the boundary layer height. It is commonly observed that the
sensible heat flux H decreases with height, while the latent heat λE increases with height. Adapted from
Stull, p.442, (1988).
fluxes by 68 W/m2 which is a 36% relative increase.
In the following step the three presented factors are qualitatively put together. A
misrepresentation of the modelled surface turbulent fluxes due to the coarse two reservoir
resolution implies that the modelled latent heat flux should be increased, whereas the
modelled sensible heat flux should be decreased in order to correspond with the realistic
surface values. An exclusion of the greenhouse term implies an increase in the both
modelled fluxes. The sum of these two corrections would therefore imply a significant
increase in the modelled sensible heat flux and a slight increase in the latent heat flux
(Table 5.1). Finally, the negative instrumentation bias means a necessary increase in both
turbulent heat fluxes from the observations. Consequently, a consistent consideration of
the three factors would most likely lead towards a better agreement of the latent heat
fluxes and also to a significantly better match of the sensible heat fluxes. The overall
implications of the qualitative analysis are presented in Table 5.1. The qualitative analysis,
which outlays some potential improvements in the framework thus only strengthens the
argument of the framework’s potential applicability in modelling the surface turbulent
fluxes and in that way confirms our second hypothesis.
Another notable demonstration of the the framework’s applicability is its ability to
give reasonable first order estimates of the transport within the CBL. By relating the
vertical exchange velocity w to the convective velocity scale w∗ (Eqs. 4.1 and 4.2) we
have connected two distinct boundary layer concepts, namely the dynamic boundary layer
and the EBL. The modelled convective scale velocity is therefore not just an additional
61
Uncorrected values
Vertical uniformity
Instrument bias
No GHG
Hobs
109.0 W/m2
/
↑
/
Hmodel
77.1 W/m2
↑
/
↑↑
λEobs
96.4 W/m2
/
↑↑
/
λEmodel
110.7 W/m2
↓
/
↑↑
Table 5.1: The table presents the identified factors that have affected the bias of the modelled surface
turbulent heat fluxes. The effects are presented with arrows. An upward facing arrow suggests a positive
correction of the flux, wheareas the downward facing arrow suggests a negative correction. The number
of arrows demonstrates the assumed magnitude of a recommended correction of the variable. Overall,
we can see that the modelled and the observed λE and H should probably be increased. The proposed
corrections would presumably lead to a better agreement between the observations and the model.
modelling feature. It is also an affirmation that the two boundary layer concepts are not
mutually exclusive but rather complementary.
5.1.3
The Changing Climate Experiments
A changing climate sensitivity provides insight in the model’s response to an enhanced
greenhouse effect. The projected sensitivity responses which were presented in the results
chapter are generally comparable with the trends of the more complex climate models.
There are however considerable differences in the magnitudes of the climate projection
trends. In this subsection we further discuss the interpretation of the results and explain
their implications for the climate system.
The analysis of the model’s surface energy balance sensitivity shows that the increases
in the outgoing surface longwave radiation RL,surf and latent heat λE are balanced by
equal decreases in the atmospheric longwave radiation RL,atm and sensible heat H. Therefore, on contrary to the referenced models, our climate experiments suggests that there is
no interaction between the longwave components and the turbulent heat fluxes responses.
Increases in the surface longwave radiation RL,surf and latent heat λE are balanced by
the equal decreases in the atmospheric longwave radiation RL,atm and sensible heat H.
The reason behind an inconsistency between the referenced and modelled surface energy
balance sensitivity analyses lies in the simplistic longwave radiation transfer of our model.
The sensitivity analysis showed that the RL,surf response of the model equals a change in
the incoming longwave radiation: ∆RL,surf = 43 ∆τ · RL,atm Consequently, even though the
equation for a difference in the surface energy budget at different climatological conditions:
∆Rnet = ∆H + ∆λE (Boer, 1993) holds also for our model, the model does not consider a
possible energy transfer between the longwave and turbulent heat fluxes because the longwave interdependence allows no changes in the net radiation at the surface (∆Rnet = 0).
62
An oversimplified parametrization of the longwave radiation also explains the overestimation of the sensible heat flux response to a surface warming. The model also projects a
significantly lower climate temperature sensitivity Λ = 0.21°C/W when compared to the
mean value of the equilibrium climate sensitivity range value Λ = 0.81°C/W (Collins et
al., 2013). Nonetheless, we should not jump to conclusions about the model’s performance.
Naturally, the simple framework does not incorporate the positive climate feedbacks, such
as the cloud feedback, the vapor feedback and the ice-albedo feedback which are essential
features to be considered in a comprehensive climate sensitivity analysis. Therefore, a
smaller climate sensitivity of the model, when compared to the sensitivities of the state
of the art GCMs, should be expected. As a matter of fact, the sensitivity analysis of
the model closely resembles the climate sensitivity due to the greenhouse effect without
climate feedbacks: Λ = 0.27◦ C/W , which is given in the IPCC’s Third Assessment Report (IPCC, 2001). Exclusion of the positive feedbacks also explains why the modelled
sensitivity perfectly fits a linear function (Fig. 4.11).
Sensitivity analysis of the mean diurnal convective transport showed the best agreement with the reported values of the mean global convective transport. The results
suggest strong global warming implications on the convective systems. Would vertical
rates of transport really be decreased by ∼ −6% ◦ C, as suggested by our model, that
would imply an approximate reduction of the vertical transport by about 20 % by the
end of the century for the RCP 8.5 climate scenario (Collins et al., 2013). The corresponding weakening of the Hadley cell could seriously alter the patterns of heat transport
to the mid-latitudes. Finally, we should briefly discuss the implications of an enhanced
evaporation on the water cycle. Normally one would expect that higher rates of evaporation would also imply more precipitation. That may be true on a global scale, however,
the water cycle is a complex non-local system implying that a steady state assumption
between precipitation and evaporation does not hold for the local systems such as ours.
An enhanced evaporation alone therefore does not provide a full picture of the effects on
the water cycle at a particular site. While an enhanced evaporation can lead to the drying
of the soils and vegetation, the actual state of the modules clearly also depends on the
local changes in precipitation (Collins et al., 2013).
63
5.2
Limitations
Our simple framework is evidently subjected to a number of limitations, some of which
were already briefly outlined in the previous sections and in the results chapter. Here we
present the most crucial limitations and hence outline the possible future improvements.
5.2.1
The Two Reservoir Representation of the Atmosphere
Even though an introduction of the EBL reservoir was an improvement over the previous MPP frameworks with a single atmospheric reservoir (Kleidon, Renner and Porada,
2014), it still provides a coarse resolution in the model. As it was explained in the previous section, the single EBL reservoir imposes a uniform heat transfer in the system, thus
making the framework generally applicable only for the certain conditions that resemble
the presence of a well-mixed CBL. Even then, with an assumed EBL depth of ∼ 5 km we
are representing a vertical domain that is more than twice the size of the typical CBL
vertical domain (1500-2000m) (Stull, p.460, 1988) Commonly observed conditions of a
non-linear vertical temperature profile and temperature inversions in certain parts of the
boundary layer therefore cannot be reproduced with this framework. A coarse resolution
is even more problematic in the TOA reservoir, the energetic reservoir that aggregates
the properties of an entire upper troposphere and stratosphere, which are thermodynamically very diverse, with a single reference temperature TA . That temperature is together
with the surface temperature TS regulating the entire longwave exchange in the system.
Moreover, the top of the boundary layer temperature TT BL is directly coupled with the
TOA temperature, therefore TA additionally also co-regulates the surface turbulent heat
transport. In reality however, the stratospheric temperatures and the temperatures in
the upper troposphere, which are represented with the TA do not play a role in the landatmosphere exchange. This unrealistic coupling of the surface and the TOA reservoir
is clearly reflected in the diurnal cycle and an overestimated amplitude of the TA (Fig.
4.9b). That also implies a high variability and an overestimation of the atmospheric
longwave radiation RL,atm around noon and its underestimation in the morning and late
afternoon. Atmospheric radiation is an important source of energy for the surface reservoir, therefore the overestimated variability of the TOA reservoir temperature also affects
the surface temperature (Fig. 4.10). In that way the coupling explains an unrealistic
diurnal variability of the TS .
Besides this misrepresentation of the temporal variability in the temperatures, the
model also overestimates the surface temperature magnitudes with the modelled mean
temperature TS approximately 14◦ C higher than the observed mean value at Hyytiälä.
The higher TS as well as the overestimated surface longwave components are a consequence of the opaque atmosphere assumption. With this assumption we assume no direct
longwave radiation release from the surface into space. The energy that should be emitted into space is instead absorbed in the atmosphere with a significant part then radiated
back to the surface. The opaque atmosphere therefore overestimates the longwave radia-
64
tion fluxes at the surface, which results in a considerably warmer surface. A back of the
envelope calculation of the surface energy balance, assuming that∼ 20% of the surface
radiation RL,surf is directly radiated into space (Trenberth et al., 2009; Staley and Jurica,
1972), shows a reduction in the surface temperature by ∼ 13◦ C, which brings the mean
surface temperature very close to the observed mean value.
The two reservoir representation of the atmosphere also does not explicitly model the
clouds which limits the framework’s applicability to the relatively cloudless meteorological
conditions. The effect of clouds is actually implicitly included in the framework because
the clouds decrease the measured solar radiation at the surface (1−ϕ)Rs , which is used as
a forcing input in the model. However, this representation proves to be insufficient. As we
know, the model projections are made from the current state of the forcing variables, thus
ignoring the previous states of the system. In that way the model disregards the inertial
processes in the system that for example sustain the temperature gradients. Whereas the
presence of clouds decreases the absorbed solar radiation at the surface, this does not
directly imply an immediate temperature difference decrease in the system, as suggested
by our framework. This example clearly demonstrates a significant deficiency of the energy
equilibrium calculations without referring to the state of the system in the previous time
step.
5.2.2
Soil and Vegetation Modules
A significant effort was invested in the soil and vegetation modules design, which is
reflected in the relatively advanced 4-layer soil heat flux module, a 3-layer soil moisture
module and an isothermal canopy vegetation module (appendix A). The modules are
relatively complex when compared with the considerably coarser atmospheric heat engine
framework. That raises the question whether the additional complexity in the modules
considerably improves the quality of our framework, or should the invested time be more
rationally spent for other model improvements? The results, presented in the appendix
A, undoubtedly show that the modules capture the observed moisture and soil heat flux
dynamics reasonably well. However, the precision of the modules’ outputs is actually
much less detrimental for the physical consistency of the framework than some other
issues, which could be improved, for example the opaque atmosphere assumption. In
addition, a possible implementation of the canopy heat storage has been proven to be
much more essential to improve the precision of the model than the additional reservoirs
or layers in the soil. The obtained results and gained insights in the application of the
framework have thus led to a redefinition of the priorities of the framework components,
which should be considered in the future.
65
5.2.3
The EBL Heat Storage
Energetic partitioning of the atmosphere with an introduction of the EBL heat storage,
which was proposed by Kleidon and Renner (2014), is a fundamentally new concept
in the MPP heat engine model. The extension transforms a static representation of
the convective heat transfer, occurring between the surface and atmospheric reservoirs
(Fig. 1.2), into a diurnally variable process, which interactively responds to the energetic
state of the near-surface atmosphere (Fig. 1.3). An analytical calculation of the storage
term together with the heat storage calculated from the radiosonde measurements agree
reasonably well with the model outputs (Fig. 4.7), thus suggesting that the energetic
partitioning of the atmosphere is physically consistent. However, a wide scatter also
demonstrates the difficulty of inferring the EBL heat storage from measurements (Fig.
4.6). In the methodology chapter, it was suggested that the main source of error might
stem from the role of synoptic weather systems, which are independent of the local surface
heating. These changes lead to a quick drop/increase in the surface pressure and vertical
temperature profile, which can have a dominating effect on the calculated sensible heat
storage HaSH (Eq. 3.4). We proposed a solution by introducing a fixed pressure thickness
of the EBL ∆p, thus effectively conserving the heat capacity of the air in the column, which
removes the noise effects of the surface pressure tendency. Nevertheless, even with a fixed
pressure thickness of the column, the system is still subjected to the horizontal transport
of heat and moisture, which are changing the energy properties of the atmosphere at a
measuring site. Therefore, our solution does not fully eliminate the effects of synoptic
changes. Moreover, it should be noted that the pressure tendency term is also a result
of a local heating. By eliminating the second term in Eq. 3.4 the surface heating effect
is unavoidably eliminated and the storage term can become either underestimated or
overestimated. Calculation of the storage from radiosonde measurements is for that reason
alone, besides the errors in the measurements, always subjected to a mistake, either by
an inadvertent inclusion of the synoptically associated pressure tendency coming from
a horizontal heat transport or by an exclusion of the local pressure tendency. These
points stress the importance of including the horizontal heat advection in the EBL heat
storage calculation in order to improve the system’s projections on a daily basis. The
current implementation of the EBL is only applicable when the local heating dominates,
a good example being the calm and cloudless sunny days. Furthermore, the radiosonde
measurements do not necessarily reflect the local heat storage at the observation site, as
they are normally being horizontally transported by the wind. In that way, during their
ascent to the 400 hPa, the radiosondes can measure the atmospheric properties over the
surface types which have little or no resemblance to the surface at the site of their release.
Another problem in the current EBL heat storage design is an inaccurate representation of the greenhouse effect. While it is true that the water vapor in EBL does play
a significant role of an absorbent in the greenhouse effect, thus absorbing a fraction of
the longwave radiation in the system, the water vapor is also emitting longwave radiation (Held and Soden, 2000). Furthermore, the boundary layer is usually (when there is
66
no vertical temperature inversion) a net emitter of the longwave radiation (Stull, p.507,
1988), though the longwave cooling is not the dominating energetic term in the CBL.
Our representation of the greenhouse effect in the EBL disregarded the emitted longwave
radiation, thus wrongly attributing the reservoir with an additional energy input. This
mistake should be corrected by assuming a net zero greenhouse heating of the boundary
layer, which would actually simplify the calculation of the turbulent heat fluxes (Eq. 5.1).
The correction would imply an average increase in the surface turbulent fluxes by ≈ 36%.
As a consequence of the increased surface turbulent heat fluxes, the mistake does not
affect the values of the EBL heat storage because the decrease in heating, associated with
the greenhouse enhancement, would be compensated by an increased turbulent heating.
5.2.4
Power Generation and Dissipation in a Steady State
Another limitation relates to the power-dissipation steady state assumption P = D.
While turbulent convective dynamics and dissipation are indeed two inherently related
processes, they certainly are not codependent as suggested by the steady state. First of
all, the assumption is not realistic due to a chaotic nature of the turbulent processes (Stull,
p.168, 1988). Clearly the system cannot be treated in a steady state during the power
generation dominated buildup of the boundary layer in the morning and the dissipation
dominated breakdown of the boundary layer in the late afternoon (Kleidon, Renner and
Porada, 2014). Yet, the diurnal average of the system can be regarded to have reached
a steady state, particularly in the periods of a relatively constant sustenance of the convective motions (Kleidon, Renner and Porada, 2014). We briefly outline a theoretical
proposal how to circumvent the steady state assumption. We first introduce the aerodynamic friction force which determines the rate at which the power generation of the
atmospheric engine is dissipated near the surface:
Fdrag = CD · ρv 2
(5.3)
where CD is a drag coefficient, depending on the dynamic stability of the atmosphere and
v is near surface wind at the chosen reference height. This is a standard approach in the
surface friction modelling (Kleidon and Renner, 2013; Garratt, 1977). We rewrite the
friction equation into the Reynolds stress form (Stull, p.63, 1988):
CD · ρv 2 = (ρv) · w
(5.4)
This enables us to interpret the friction force as a result of a surface pull on the air
due to a momentum exchange ρv between the surface and near-surface atmosphere that
happens with an effective vertical exchange velocity w = Cd v (Kleidon and Renner, 2013).
The friction force dissipates the power near the surface at the rate that is proportional to
the near-surface wind D = Fdrag · v = ρw · v 3 (Kleidon and Renner, 2013). This parameterization of the dissipation introduces an additional unknown variable in the system of
equations, namely the near surface wind v. To obtain a determined system of equations,
67
we have to express the latter variable as a function of other variables and parameters. One
of the possible approaches to resolve this problem is application of empirical boundary
layer relations. Garratt, (1977) suggests to use the following experimental relation be2
. By using such an
tween the exchange velocity and the near-surface wind w = 0.75·v+0.067·v
10000
empirical relation, in order to express v with w we implicitly assume that the dissipation
is proportional to the intensity of convective motions in the system w∗ , an assumption
that was also made in the proposed self-regulating MPP mechanism (Ozawa et al., 2003).
A further improvement would be to relate the velocities without applying the empirical
relations. A decrease in reliance on the empirical relations was after all one of the main
advantages of the MPP framework in relation to the conventional frameworks. Such an
extension should enable us to model the surface turbulent fluxes without the steady-state
assumption, which should in theory improve the precision and physical adequacy of the
framework. On the other hand, an application of the empirical relations advantages of
our framework, namely reliance on empirical parameterizations.
5.2.5
Representation of the Water Cycle
We derived the entropy production in the system by applying the assumption of no
latent heat release in the boundary layer (Eq. 2.24). While the moisture input in the
system is supplied by the latent heat flux, the discharge of moisture from the EBL is
represented only implicitly, namely by the outgoing turbulent heat flux Jout . The Jout
accounts for the transport of the remaining sensible and latent heat from the heat engine
into the TOA reservoir. There the latent heat release is not properly addressed as the
framework does not simulate precipitation. The water supply from precipitation that
provides water for soil moisture is therefore modelled from the measured precipitation
rates. Even so, the framework’s representation of the water cycle still fails to capture the
actual circulation of water moisture in the system. As we know, the precipitation i.e., the
convective latent heat release predominantly occurs within the boundary layer (Koenings
et al., 2012). The considered role of moisture transport into the TOA reservoir is therefore
inadequate. Also, the water cycle generally cannot be treated as a closed local system.
The precipitation events are especially in the high latitudes normally dominated by the
advected convective systems and weather fronts, which would require an introduction
of the horizontal moisture transport. Surprisingly, the deficiencies in the water cycle
representation did not decisively affect the comparative EBL heat storage analysis as
the modelled EBL heat storage reproduced the diurnal variability of the heat storage
calculated from the radiosonde data reasonably well. This is probably a consequence of a
relatively dry measuring period with only three notable precipitation events thus making
the representation of the water cycle irrelevant in the most model runs.
68
5.3
Prospective Applications of the MPP in the Complex Models
We conclude the discussion by outlining the possible advantages of applying the MPP
limit in the more complex models. While the conventional models have a capability
to evaluate the thermodynamic limits in the system, these limits are not used as thermodynamic constraints of the system’s dynamics (Kleidon et al., 2006). The research
that we have conducted demonstrates that the MPP limit can be employed to roughly
reproduce the dynamics of the energetic land-atmosphere interactions. The proposed atmospheric heat engine framework therefore has a potential of becoming a useful concept
in atmospheric modelling science. Yet, there is a long way to that point with numerous
challenges on its way. The first prerequisite would be to improve the simple conceptual
thermodynamic limit models by adding important processes to the framework, which are
at this point neglected and and should improve the framework’s precision and extend
its applicability to different atmospheric conditions. However, even an improved version
of the atmospheric heat engine with an additional degree of complexity would still have
to be subjected to a thorough performance analysis. Such an analysis would probably
be analogous to our comparative analysis between the model and observations. If such a
framework proved itself adequate to model the energetic dynamics in the land-atmosphere
system, this could provide an alternative approach for the modelling of land-atmosphere
exchange. Such an alternative approach could improve the precision of the current GCMs,
since these models largely rely on the empirical functions from the similarity theory. While
it is true that the empirical functions were derived from numerous detailed field studies,
these functions do not explain the fundamental nature of the turbulent processes. It is
also clear that the empirical functions, derived from particular sites do not apply equally
well to the varieous sites with markedly different systemic properties (Stull, p.348, 1988).
The atmospheric heat engine framework on the other hand does not require the empirical
functions of the static and dynamic stability in the atmosphere to model the exchange
processes, which are in our model represented by w. Rather, in the presented framework, the state of the exchange processes in the system is directly inferred from the MPP
generation limit. Finally, if the tendency to maximize power generation will eventually
be proven as an inherent characteristic of the system, that would be a tremendous leap
forward in the general understanding of the land-atmosphere interactions.
69
Chapter 6
Conclusions
A simple energy equilibrium framework has been devised to test the applicability of
the MPP limit in land-atmosphere interactions on the diurnal time scales. The underlying working hypothesis of our research was that we can describe the processes of vertical
turbulent heat transport and convective motions by treating the atmosphere as an idealized heat engine, operating at its thermodynamic limit. Such an atmospheric heat engine
strives to maximize power generation in the atmosphere by converting a fraction of the
vertically transported turbulent heat flux into convective motions. The convective heat
transport tends to take place in the lower troposphere, a part of the atmosphere that is
energetically influenced by the turbulent heat fluxes from the surface. For that purpose
we have introduced an energetic boundary layer (EBL) which constitutes the operational
domain of the atmospheric heat engine. Quantification of energy and entropy production
in the EBL has enabled us to make the first order estimates of the surface turbulent heat
fluxes E & λE and convective transport w∗ in the system. A detailed comparative analysis
of the model results with the measurements from Hyytiälä has demonstrated the framework’s ability to decently project the diurnal cycle of the surface turbulent heat fluxes
and convective transport. The average magnitudes of the projected fluxes have shown
a lesser resemblance with the observations. Still, a clear correlation between the model
and the observations has proven that there is a consistent representation of the energetic
dynamics in the local atmospheric system. Moreover, we have proposed the solutions to
the identified flaws in our framework that have affected the precision of the projections,
and have provided reasoning that an implementation of these solutions would bring the
observations and the model to a better agreement.
Reasonable representation of land-atmosphere interactions with our model has demonstrated that the atmospheric system indeed operates near its thermodynamic state of
maximum power generation. In that way has the comparative analysis justified the most
important working hypothesis of our research, i.e., treating the atmosphere as an idealized
maximum power generating heat engine. Yet, even though the results suggest that the
system’s dynamics can be modelled with the MPP approach, this in itself does not prove
that the system seeks such a state. In order to fully justify such an assertion, the empir-
70
ical analysis of the model and observations has to be complemented with a fundamental
physical theory which would explain the tendency of the system to maximize power generation. Our contribution to that aspect of research was the conceptual illustration of the
self-regulating MPP mechanism which was adapted from Ozawa et al., (2003) and Kleidon
and Renner, (2013). While the illustration has provided a plausible conceptual reasoning
for the MPP limit, the existence of such a self-regulating mechanism remains uncertain.
That is because the more particular applications of disequilibrium thermodynamics, such
as the MPP limit, constitute a relatively new and unexplored field of research. Therefore,
further evaluations are needed to test the applicability and scope of validity of the MPP
limits. Crucial prerequisites for a more detailed and rigorous study would be to improve
the precision and physical consistency of the alternative models which would extend its
applicability to a wider range of observed meteorological conditions. We have identified and outlined the most fundamental improvements for out framework. A multiple
heat storage framework would increase the vertical resolution of the model and extent its
applicability to non-convective situations. An explicit representation of the dissipative
processes would improve the consistency of the framework during the periods of the day
when power generation and dissipation are not in a steady state. Finally, the local water
cycle extension would broaden the physical consistency of the framework to the situations
of wet condensation in the EBL.
The contribution of this research is twofold in its nature. First of all, the demonstration of the model’s capabilities in estimating the surface turbulent heat fluxes puts the
framework into a perspective of becoming an practically applicable modelling tool.
With an improved precision and reliability, the MPP models could become useful for
providing the estimates of the surface turbulent fluxes on the local spatial scales. Framework’s simplicity that is best captured by the reliance on generally accessible input data,
which can be obtained from the satellite measurements, could make these models suitable
for estimating the surface turbulent fluxes worldwide, including the remote and underdeveloped regions which lack the field measurements. Secondly, the research has identified
and structured the future potential uses of the MPP limits in the more complex models.
We have argued that the MPP approach could potentially complement the conventional
GCMs by adding an additional physical constraint to the system, thus reducing their
reliance on the empirical parametrization of the fluxes. A more advanced application of
the thermodynamic limits could in that way potentially lead to an improved precision of
the conventional GCMs. Furthermore, the alternative approach could facilitate a more
comprehensive understanding of the fundamental principles of land-atmosphere interactions. At the same time, we have also pointed out some of the challenges that lie ahead
before these potentials could materialize. From that perspective we view this thesis as a
study that has proven the applicability of the alternative MPP approach and has outlined
its prospectives which is why it may serve as a reference guide for future research.
71
Appendix A
Soil and Vegetation Modules
In this appendix we provide a more detailed description of the soil and vegetation
modules design. We also present the model outputs and compare them with observations
from Hyytiälä.
A.1
A.1.1
Soil Heat Flux
Model Implementation
Heat transfer and temperature in the soil were calculated from the Fourier’s law for
heat conduction (Eq. 2.30) and the energy balance equation for each of the four soil
layers over a selected period of ∆t = 30 min (Eq. 2.32). These equations are solved
numerically and are therefore presented in the finite difference form (Eq. A.1). The
system of equations is solved with a numerically stable forward Euler method, where ∆z
stands for the thickness of a single soil layer, ρ refers to the density of the soil and cs is
the specific soil heat capacity:
ρs,i · cs,i · (Ts,i (t + ∆t) − Ts,i (t))
Qg i−17→i − Qg i7→i+1
=
(A.1)
∆z
∆t
The soil is partitioned into 4 layers so the boundary conditions have to be applied on
the top layer and the fourth layer of the soil in the framework. For the top soil layer,
which is in contact with the surface, the Fourier’s equation represents the conduction
between the surface TS and the top soil layer Ts,1 . Conduction between the surface and
the middle part of the top layer implies that we calculate the heat transfer only over
one half of the layer depth (Eq. A.2). This equation characterizes the soil heat flux as
proportional to the current value of thermal conductivity ks and temperature difference
between the surface TS and the top soil layer Ts,1 . For the purpose of calculating the soil
heat flux we used the near surface temperature from the observations and not the model
output!
72
Qg (t + ∆t) = −
k · (TS (t + ∆t) − Ts,1 (t + ∆t))
∆z
2
(A.2)
For the fourth layer a no flux boundary condition is applied, thus assuming that only
the top four soil layers exhibit a noticeable energy variations on the diurnal time scales
(Garratt, p.117, 1992) (Eq. A.3).
F37→4 − 0
Ts,4 (t + ∆t) − Ts,4 (t)
=
(A.3)
∆z
∆t
In that way we have obtained a system of four equations for soil energy balance with
five unknowns, i.e. the soil temperatures in the layers Ts,i . Solving the system of these
equations allows us to calculate the soil heat flux (Eq. A.2).
A.1.2
Impact of Moisture on the Conductive Properties of the
Soil
Soil moisture storage (also the degree of saturation) θ has an indirect effect on the
convective properties of the soil which is why the soil heat flux calculation has been
coupled with the calculation of the moisture storage in the soil. One can imagine soil as a
multitude of differently shaped soil particles in contact with each other. Different shapes
of the particles allow that a large portion of the soil is usually occupied by pores which
can be filled with water or air. Fraction of the pores that are filled with water determines
the moisture storage within the soil. The degree of saturation changes the density ρs ,
thermal diffusivity κs and heat capacity of the soil cs . In this subsection we present
the implementation of the of soil density, specific soil heat capacity and conductivity
dependencies on the moisture storage.
A.1.2.1
Soil density
Soil density ρs is increased by an additional input of moisture. As more water is
added to the soil, the air with negligible mass is displaced by water. A new parameter,
the soil dry density ρd is introduced. Dry density of the soil stands for the density of a
soil without any moisture. This enables us to calculate soil density as a function of the
degree of saturation (Eq. A.4):
W
· θsat · ρw
ρs = ρdry +
Wmax.
(A.4)
where W represents the height of the water column in the bucket with a maximum
capacity of Wmax . The proportion of the maximum bucket capacity in relation to the soil
layer depth defines the moisture saturation value of that layer θsat = Wmax /dz.
73
A.1.2.2
Heat capacity
Soil is usually not chemically homogeneous. It consists of different chemical compounds with their corresponding fractions. Each of these compounds has its own heat
capacity. One can assume that these compounds are evenly distributed throughout the
soil. Therefore the aggregate soil heat capacity must be calculated as a sum of capacities
from individual compounds, multiplied by the mass fraction of each compound xi :
cs = x1 c1 + x2 c2 + ... + xi ci ; where : x1 + x2 + ... + xi = 1
(A.5)
We can further simplify the formula by assuming that the soil can be described as
a sum of mineral compounds cq , organic compounds co and water cw , with compounds
within a particular group having the same heat capacity (De Vries, 1975):
W
(A.6)
· θsat · cw
Wmax
Fractional parameter q stands for the proportion of mineral compounds in the soil.
cs = (1 − θsat ) q · cq + (1 − q) · (1 − θsat ) · co +
A.1.2.3
Thermal conductivity
Numerous soil scientists have been working on the development of semi-empirical engineering methods, which try to capture the dynamics of the soils’ conductive properties.
“Thermal properties of soils” by Farouki, (1986) provides an extensive overview and evaluation of these methods for a wide variety of soils. For the purposes of calculating thermal
conductivity as a function of moisture for the Hablic podzol, the soil type at Hyytiälä,
Farouki (1986) recommends the Johansen’s method due to its reliability and satisfactory
precision. The following equations that constitute the Johansen’s method (Eqs. A.7 A.14) are all adapted from Farouki, (1986). Johansen’s method to calculate thermal conductivity at a partial saturation of the soil is based on the concept of the Kersten’s number
Ke , which is basically an interpolation between the conductivities of the saturated ks,sat
and dry state ks,dry (Eq. A.7).
Ke =
ks − ks,dry
ks,sat − ks,dry
(A.7)
By knowing the Kersten number Ke we can calculate the conductivity as:
ks = (ks,sat − ks,dry ) · Ke + ks,dry
(A.8)
Based on the experimental data, Johansen has derived the following relationships
between theKe and the degree of saturation θ = WW
for a fine soil (Farouki, p.113,
max
1986):
W
Ke = log
+ 1.0
Wmax
(A.9)
74
Furthermore, the author has also formulated the semi-empirical equations for determining the dry and saturated state conductivities ks,dry and ks,sat for different types of
soils. Here we present only the relations that are applicable to the unfrozen natural soils,
which correspond with the properties of the Hablic podzol. First of all, we calculate the
thermal conductivity of a dry soil ks,dry with the following empirical relation (Farouki,
p.112, 1986):
ks,dry =
0.135 · ρdry + 64.7
2700 − 0.347 · ρdry
(A.10)
This is followed by a calculation of the effective thermal conductivity of the solids
ks,solid , which characterizes the thermal properties of the soil particles, i.e soil without
pores, depending on the content of minerals (quartz) and organic matter (Eq. A.11).
ks,solid = kqq · ko1−q
(A.11)
Finally, we need to obtain the thermal conductivity of the soil at a saturated state. Johansen proposes an empirical relation, where the conductivity is calculated as a geometric
mean of the conductivity of the solids ks,solid and water kw , with each term exponentiated
at its respective volume faction (1 − θsat ):
1−θsat
ks,sat = ks,solid
· kwθsat
(A.12)
With the known information of dry soil density ρdry , moisture saturation θsat , the
content of quartz q and moisture storage W we can calculate all of the above mentioned
parameters and use them to evaluate the thermal conductivity of the soil ks as a function
of soil moisture storage (Eq. A.8).
A.1.3
Soil Heat Flux Results
We compare the modelled and measured soil heat flux Qg at Hyytiälä from 13/07/2010
to 10/08/2010. This is followed by an analysis of a diurnally averaged soil heat flux. Figure
A.1a depicts a continuous time-series of the modelled and measured soil heat flux. The
time step on the x axis represents a 30 minute interval between two consecutive values of
the flux. The positive values of the heat flux show the net heat storage in the soil, while
the negative values imply a net heat release from the soil. We can see from the Fig. A.1a
that the model captures the diurnal cycles reasonable well, yet on the other hand also
clearly overestimates the amplitude of the flux. Moreover, it can also be observed that
the measured heat flux rarely drops bellow zero. The observations therefore imply that
the soil is a net absorber of the heat, which does not make sense over longer time periods
(Garratt, p.118, 1992), when the daily average of the soil heat flux should be ∼ Qg = 0.
For that reason, we intentionally did not try to calibrate the soil heat flux according to the
measurements. We also intentionally calibrated the flux in a way that we have obtained
a larger amplitude response of the soil. In that way, we tried to partially compensate for
75
Figure A.1: Soil heat flux Qg model analysis. Fig. A.1a depicts the time-series of the observed and
measured soil heat flux. Time step on the x axis represents the 30 minute intervals of the series. Fig.
A.1b portrays an averaged diurnal cycle of the modelled soil heat flux. The figure shows that the soil
becomes a net absorber of the heat around 7am, which is approximately one hour after a positive increase
of the near surface temperature. A similar time delay is observed in the late afternoon when the soil heat
flux becomes a net emitter of heat after 7 pm.
the negligence of the canopy heat storage, which is in principle a more important term in
the surface energy balance. Figure A.1b depicts a diurnally averaged modelled soil heat
flux. By comparing the diurnal variability of the flux with the diurnal variability of the
observed near surface temperature (Fig. 4.9a), we can observe a positive time delay of
the soil heat flux of approximately one hour, which is a reasonable estimate for the delay
of a soil heat response in the forest (Garratt, p.236, 1992).
76
A.2
A.2.1
Soil Moisture
Groundwater flow
Groundwater flow is based on the Darcy’s law, a commonly used approach in soil
moisture modeling (Garratt, p.138, 1992) (Eq. A.13). The two variables that determine
the magnitude of the groundwater flow are the hydraulic conductivity K(W ) and the
hydraulic pressure head ψ. Both are specified with empirical functions for various soil
types (Clapp and Hornberger, 1978).
∂(ψi + z)
(A.13)
∂z
Hydraulic conductivity K(W ) is proportional to the hydraulic conductivity value at
the saturation Kθsat , multiplied by soil moisture storage θ which is exponentiated with an
empirical parameter β (Eq. A.14). Hydraulic pressure head ψ has an analogous functional
form (Eq. A.15).
Fg,i = −ρw · K(Wi ) ·
K(Wi ) = Kθsat · (
ψi = ψsat (
A.2.2
Wi 2β+3
)
Wmax,i
Wi −β
)
Wmax,i
(A.14)
(A.15)
Water Availability in the Bare Soil
Total evapotranspiration was partitioned into the bare soil evaporation and evapotranspiration of the vegetation (Eq. 2.41). The fractions reflect the surface type proportions
at the observation site that is covered by vegetation and bare soil, a method that is applied
in the ECMWF model (IFS Model cycle - Cy40r1, 2013). The fraction estimates, which
were based on the satellite imagery of the measuring site, are as follows; 30 % of the total
evapotranspiration at the Hyytiälä site is attributed to the bare soil evaporation λE0 . The
residual 70% is attributed to the evapotranspiration of the vegetation which is described
in Eq. 2.40. The parameter that relates bare soil evaporation at the surface with soil
moisture storage is water availability fw (Kleidon et al., 2014). Water availability, with its
value ranging from 0 to 1, depends on the specific humidity gradient between the surface
and top of EBL q(TS ) − qsat (TT BL ). Specific humidity at the surface is calculated as a
saturated specific humidity, which is a function of surface temperature multiplied by the
relative humidity rh at the surface of the bare soil (Eq. A.16). The expression was derived
by equating the model parameterization of the bare soil latent heat flux (Eq. 2.29) with
the “Monin-Obukhov parameterization” (Eq. 2.27).
fw =
rh · qsat (TS ) − qsat (TT BL )
qsat (TS ) − qsat (TT BL )
(A.16)
77
Relative humidity at the surface rh is calculated from an empirical function by Clapp
and Hornberger and relates the relative humidity to the soil moisture storage in the top
(Garratt, p.138, 1994).
layer of the soil θ = WW
max
−g · |ψsat | · ( WW
)−β
max
rh = exp(
)
Rv · TS
Finally, the saturated specific humidity is calculated as:
qsat (T ) =
A.2.3
0.622 · esat (T )
p − 0.378 · esat (T )
(A.17)
(A.18)
Soil Moisture Storage Results
Analysis of the soil moisture in the top soil horizon (top 7 cm of the soil) which is
detrimental for the bare soil evaporation was conducted by comparing the modelled water
column height with the observed water column height (Fig. A.2). Maximum field capacity of the top soil was determined at 17.5 cm. For the purpose of qualitative analysis a
water limited regime was determined, which was defined as the water level below which
the water availability parameter fw drops under 0.8. The water limited regime boundary
was calculated from the equations A.16 and A.17 and equals ∼ 4 cm. We can see from
Fig. A.2. that the modelled moisture storage closely matches the observations from 13th
of July to 29th of July. In the period between 29th of July and 4th of August the model
overestimates the drying of the top soil. On contrary to the observations, which demonstrate a halt in the drying, the model projects an intensification of the drying. Further
on the precipitation event on the 4th of August increases the moisture storage of both
time series. Finally, the precipitation event on the 8th of August noticeably increases
the modelled moisture storage, which returns back to the observed values. Besides the
aforementioned overestimation in the drying of the soil this latter disproportional increase
in the moisture storage cleary indicates that the model is oversensitive under certain conditions. A detailed analysis of this problem was not performed, since the soil module was
not in the central focus of the research. Still, a possible explanation for an overestimated
drying of the top soil comes from the fact that the modelled moisture storage never drops
beyond the water limited regime. This means that the evapotranspiration was not significantly inhibited by a drop in soil moisture storage, since it never dropped below the water
limited regime boundary of fw = 0.8. Such a weak response of the water availability to
the moisture storage in the model suggests that the water limited regime in the system
probably occurs earlier, at higher storage values.
78
Figure A.2: Soil moisture storage analysis. The figure portrays a comparison of the observed and
modelled heights of the water column in the top soil layer. The figure also shows a maximum field
capacity of 17.5 cm and the upper boundary of the water limited regime at 4 cm. Evaporation of the
bare soil, which is represented by the water availability parameter fw , drops significantly as the soil is
being dried below the water limited regime boundary.
79
A.3
Vegetation
In this section of the appendix we present a more detailed implementation of the
stomatal regulation of evapotranspiration. This regulation is represented with a single
stomatal resistance parameter that represents the functioning of the entire ecosystem.
Implementation of the stomatal resistance is adapted from the Jarvis model implementation in the ECMWF operational model (IFS Model cycle - Cy40r1, 2013). Stomatal
resistance in the model is proportional to the minimum stomatal conductance rmin that
depends on the vegetation type, over the leaf area index LAI. The selected vegetation
type for Hyytiälä was the evergreen needleaf forest with rmin = 250 and LAI = 6 (IFS
Model cycle - Cy40r1, 2013). In addition, the stomatal resistance is temporally varying
according to the empirical functions of the solar radiation at the surface (1−ϕ)Rs , average
root water θ and the atmospheric humidity deficit Da :
rmin
· f1 ((1 − ϕ)Rs ) · f2 (θ) · f3 (Da )
(A.19)
LAI
The inverse of the solar radiation effect on the stomatal resistance f1 is defined as:
rc =
b · (1 − ϕ)Rs + c
= min 1,
a · (b · (1 − ϕ)Rs + 1)
"
f1 (Rs )−1
#
(A.20)
where a, b and c are empirical parameters, dependent on the vegetation type. f1 is increasing with solar radiation, thus implying a higher stomatal resistance of the vegetation
that is exposed to solar radiation. The inverse of the soil moisture saturation function f2 ,
which characterizes the stomatal resistance dependence on the average root water storage
θ, is a piecewise function of three different regimes (Eq. A.21). The first, permanent
wilting point regime corresponds with the state of very scarce moisture in the soil θpwp .
In that state, the plants completely close the stomata, therefore rc → ∞ and plants do
not allow any transpiration. The second regime refers to the state of a saturated soil
θsat . Lastly, the inverse moisture saturation function in the intermediate regime is just an
interpolation between the other “extremal” regimes.
¯
f2 (θ)
−1
=



0


; θ̄ < θpwp


1
; θ̄ > θsat
θ̄−θpwp
;
 θsat −θpwp
θpwp ≤ θ ≤ θsat
(A.21)
The average root water θ is a sum of the soil moisture storage multiplied by the root
fraction in each of the three soil layers:
θ̄ =
X
rooti · θi
(A.22)
Finally, the stomatal resistance is also dependent on atmospheric humidity deficit,
a measure of moisture in the air at the canopy Da = esat (1 − rh ). The dependence is
exponential (Eq. A.23), meaning a higher stomatal resistance at greater humidity deficit.
80
In that way the plants restrict water loss under dry ambient conditions (Farquhar et
al., 1980). The exponential parameter gd depends on the vegetation type and equals
0.03 hP a−1 in our model.
f3 (Da )−1 = exp(−gd · Da )
(A.23)
81
Appendix B
Future Recommendations
Some recommendations for framework improvements were already presented in the
implications and limitations sections. Amongst other recommendations, we have proposed an inclusion of the horizontal heat and moisture advection, an elimination of the
greenhouse radiation in the EBL, an introduction of the atmospheric longwave release
into space and an explicit modelling of the surface friction dissipation. These are all
limitations that should have been resolved in order to apply the framework in the more
complex models. We propose two structural improvements of the framework for future
work. Firstly, we build on the critique of the current water cycle representation in the
model, thus presenting a design of an explicit water cycle implementation in the model.
This is followed by an illustration of an improved vertical resolution of the model.
B.1
The EBL with Water Cycle
We begin the water cycle extension by introducing an additional moisture reservoir in
the atmosphere ΘEBL , which represents the moisture storage in the atmosphere. The EBL
moisture storage term is a function of moisture input from the surface latent heat flux
and moisture output by precipitation λP rec, which is assumed to take place entirely in
the EBL (Koenings et al., 2012) (Eq. B.1). Horizontal moisture transport in not included
in this partial extension of the water cycle.
ΘEBL = λE − λP rec
(B.1)
Implementation of the EBL moisture storage also affects the energy balance equation
of the sensible heat storage. The latent heat is released upon condensation in the EBL,
thus warming up the EBL (Eq. B.2).
Ha = H + D − P + λP rec − Jout
(B.2)
Other terms in the EBL heat storage equation remain the same as in our framework (Eq.
3.2), except for the greenhouse radiation in the EBL, which is now removed from the
82
expression. The proposed water cycle also implies a change in the nature of the outgoing
heat flux Jout . While the outgoing heat flux in the current framework represents both
sensible and latent heat transport in the TOA reservoir, the Jout term in the proposed
extension transfers only the remaining sensible heat flux from the atmospheric heat engine. This is undoubtedly a physically more consistent representation of the vertical heat
transport in the system. Finally, we should note that this extension does not provide a
complete representation of the water cycle. First of all, by excluding the horizontal moisture transport we only capture the local water cycle, which can be an adequate concept
only for the some systems. The proposal also lacks implementation of the precipitation
trigger mechanism. In other words, while we have added the precipitation in the system, we still miss the physical processes that cause precipitation at first place. A simple
first-order representation of the precipitation mechanism could be implemented, by prescribing the total moisture carrying capacity of the EBL Θmax , which would depend on
the boundary layer temperature < θBL >. When the moisture in the system exceeds the
maximum capacity, the wet condensation is triggered. Such a simple approach naturally
disregards a generally complex nature of the physics of precipitation, which also depends
for example on the amount of cloud condensation nuclei and the dynamic stability of the
atmosphere (Holton, p.302, 1973). The water cycle extension scheme is presented in the
right panel of Fig. B.1.
Figure B.1: Recommended improvements in the framework. The left panel illustrates the MPP framework with multiple vertically coupled heat storage reservoirs. Vertical temperature profile in the right
hand side of the left panel indicates a possibility to apply the MPP framework to the temperature inversion conditions. The heat storage of an individual reservoir is a function of the dissipation and power
generation within the reservoir and the net difference between the heat fluxes at the top and the bottom
of the reservoir. e.g.: Ha1 = D1 − P1 + λE1 + H1 − (−H2 − λE2 )). Thickness of the reservoirs determines
the vertical resolution of the framework. The right panel demonstrates the EBL water cycle extension.
The left side of the illustration depicts the storage and release of moisture in the EBL ΘEBL , while the
right side portrays the heat storage and release in the EBL.
83
B.2
The Multiple Heat Storage Extension
A major deficiency of our model, which was observed throughout the results analysis
is a coarse vertical resolution of the framework. Such a resolution implicitly leads to the
uniform modelling of the vertical turbulent heat transport. Moreover, it also limits the
framework’s applicability to the convective boundary layer conditions. We could improve
the precision and general applicability of the model by partitioning a single EBL reservoir
into several vertically coupled heat storage reservoirs (left panel of the Fig. B.1), each at
its own respective temperature that is determined by the net exchange of the turbulent
fluxes in the system and the balance between dissipation and power generation (Eq. N.5).
It is assumed that each reservoir exchanges heat accordingly with the MPP limit. This
extension requires different parameterizations of the turbulent heat fluxes, especially for
the latent heat flux because we cannot assume the air to be saturated at the top of each
reservoir, as we have done in the single EBL (Eq. 2.28). The partitioning of the EBL
into several reservoirs requires a use of parameterizations where the sensible heat flux
is proportional to the temperature difference between the two reservoirs, and where the
latent heat flux is proportional to the difference in the specific humidity q of the reservoirs
(Eqs. B.3 and B.4)
H1 = ρcp w1 (T1 − T2 )
(B.3)
λE1 = λρw1 (q1 − q2 )
(B.4)
The reservoir heat storage Ha1 in the proposed framework is then calculated as a
difference between the turbulent heat fluxes at the bottom and the turbulent heat fluxes
at the top of the respective reservoir, in addition to the difference between the dissipation
and power generation within the reservoir.
Ha1 = D1 − P1 + λE1 + H1 − (−H2 − λE2 ))
(B.5)
Besides the turbulent fluxes, the partitioning also leads to an improved resolution of the
convective motions. As a consequence of the partitioning, a single atmospheric heat engine
is partitioned into a multitude of vertically coupled heat engines with different vertical
exchange velocities for each reservoir w. The suggested extensions affect the calculation
of the entropy production and consequently also the derivation of the respective optimum
fluxes at the MPP limit. The proposed design would therefore require solving of a far more
extensive and presumably also more complex system of equations than in the implemented
framework. Concrete derivation steps would therefore require an extensive theoretical
research and are for that reason alone beyond the scope of this thesis study.
84
Appendix C
List of Variables and Parameters
Symbol
Variable
Units
(Mean) values in the model
α
Surface albedo
/
0.93 (Betts and Ball, 1997)
HaSH
Sensible heat storage
W
m2
/
HaLH
Latent heat storage
W
m2
/
QSH
Internal energy
J
m2
/
QLH
Latent heat
J
m2
/
g
Gravitation
m
s2
9.8
e
Water vapor pressure
Pa
/
L↓
Incoming longwave radiation
W
m2
375.6
Λ
Climate sensitivity
K
W m−2
0.21
Table C.1: Overview of variables and parameters introduced in the methodology chapter.
85
Symbol
Variable
Units
(Mean) values in the model
Rs
Total absorbed solar radiation
W
m2
269.6
ϕRs
Rs absorbed in TOA
/
0.26 (Ramanathan and Vogelmann, 1997)
RL,surf
Surface outgoing longwave
W
m2
510.3
RL,atm
TOA outgoing longwave
W
m2
439.2
τ
Optical depth
/
2.036
H
Surface sensible heat flux
W
m2
77.1
λE
Surface latent heat flux
W
m2
110.7
Qg
Soil heat flux
W
m2
10.7
Jout
Outgoing convective heat flux
W
m2
??
D
Surface friction dissipation
W
m2
D=P
P
Power generation
W
m2
39.1
Ha
EBL heat storage
W
m2
90.0
Sb
EBL entropy
W
K
/
TS
Surface temperature
K
309.3
TA
TOA temperature
K
265.3
cp
Specific heat at constant pressure
W
kg·K
1004 (Holton, 1973)
Rdry
Gas constant for dry air
J
kgK
287 (Holton, 1973)
p
Pressure
N
m2
∼ 1000
ρ
Air density
kg
m3
∼ 1.2
ref
Arbitrary reference state
< θb >
EBL bulk potential temperature
K
/
h
EBL height
m
∼ 4600
J
H + λE
W
m2
187.8
σ
Stephan-Boltzman constant
W
m2 K 4
5.67 · 10−8
RL,0
Radiation constant
W
m2
71.1
Rnet
Net Radiation at the surface
W
m2
71.1
kr
Linearized radiative exchange
W
m2 K
5.66
/
Table C.2: Overview of variables and parameters introduced in the modelling framework chapter.
86
Symbol
Variable
Units
(Mean) values in the model
raV & raH
Aerodynamic resistances
/
/
w
Vertical exchange velocity
m
s
3.0 · 10−3
TT BL
Top of the EBL temperature
K
289.5
λ
Specific latent heat
J
kg
2.5 · 105
fw
Water availability
/
0.98
q(T )
Specific humidity
g
kg
/
s
Saturation vapor pressure curve
Pa
K
195.8
γ
Psychrometric constant
Pa
K
65 (Kleidon and Renner, 2013)
Table C.3: Overview of variables and parameters related to the turbulent heat fluxes parameterizations.
Symbol
Variable
Units
(Mean) values in the model
θ
Degree of saturation
/
0.1 − 0.45
Th
Throughfall
mm
s
/
W
Moisture content
mm
∼ 11
Wmax
Maximum moisture storage
mm
/
Fg
Groundwater flow
mm
s
−2.6 · 10−6
ρw
Water density
kg
m3
1000
ψ
Hydraulic pressure head
m
/
K(θ)
Hydraulic conductivity
m
s
34.6 · 10−6 (Clapp and Hornberger, 1978)
qsat (T )
Saturated specific humidity
g
kg
2.2
rh
Relative humidity
/
0−1
Rv
Gas constant for vapor
J
kg·K
462 (Holton, 1973)
esat (T )
Saturation vapor pressure
Pa
611 at 0 ◦ C (Kleidon and Renner, 2013)
root
Root fraction
/
0−1
β
/
/
4.9 (Clapp and Hornberger, 1978)
Table C.4: Overview of variables and parameters introduced in the soil part of the appendix.
87
Symbol
Variable
Value/Units
(Mean) values in the model
ρdry
Soil dry density
kg
m3
1600 (De Vries, 1975)
θsat
Moisture saturation value
m3
m3
/
Wmax
Maximum moisture storage
mm
/
co
Heat capacity of organic matter
J
kg·K
1920 (De Vries, 1975)
cq
Heat capacity of minerals
J
kg·K
840 (De Vries, 1975)
cw
Water heat capacity
J
kg·K
4182 (De Vries, 1975)
q
Mineral fraction
/
5% (Greve et al., 1998)
Ke
Kersten’s number
/
/
ks,sat
Conductivity of saturated soil
W
m·K
/
ks,dry
Conductivity of dry soil
W
m·K
/
ks,solid
Effective conductivity of solids
W
m·K
/
kw
Conductivity of water
W
m·K
/
ko
Conductivity of organic materials
W
2.0 m·K
2.0 (Farouki, 1986)
kq
Conductivity of mineral materials
W
7.7 m·K
7.7 (Farouki, 1986)
rmin
Minimum stomatal resistance
s
m
250
Da
Humidity deficit
[P a]
/
a
/
/
0.81
b
/
m2
W
0.004
c
/
/
0.05
gd
/
hP a−1
0.03
Table C.5: Overview of variables and parameters introduced in appendix A.
88
Symbol
Variable
Units
(Mean) values in the model
rc
Stomatal resistance
s
m
471
Ev
Dry canopy evapotranspiration
mm
s
77.5
E0
Bare soil evaporation
mm
s
33.2
m
Maximum canopy water storage
mm
12
LAI
Leaf area index
/
6
P rec
Precipitation intensity
mm
s
3.1 · 10−5
Cveg
Vegetation cover
/
0.7
θcap
Field capacity
m3
m3
0.4
Da
Humidity deficit
Pa
/
θpwp
Permanent wilting point
m3
m3
0.12
Table C.6: Overview of variables and parameters introduced in the vegetation part of the appendix.
Symbol
Variable
Units
(Mean) values in the model
w∗
Convective scale velocity
m
s
0.52
Fdrag
Aerodynamic friction force
N
/
Cd
Drag coefficient
/
/
ΘEBL
EBL moisture storage
W
m2
/
Table C.7: Overview of variables and parameters introduced in the appendix B and the results and
discussion chapters.
89
References
[1] Andrews, T., Forster, P. M., & Gregory, J. M. (2009). A surface energy perspective on climate change. Journal of Climate, 22(10), 2557–2570.
http://dx.doi.org/10.1175/2008JCLI2759.1
[2] Betts, A. K., & Ball, J. H. (1997). Albedo over the boreal forest. Journal of Geophysical Research, 102(D24), 28901–28909. http://dx.doi.org/10.1029/96JD03876
[3] Betts, A. K., & Ridgway, W. (1989). Climatic equilibrium of the atmospheric convective boundary layer over a tropical ocean. Journal of the Atmospheric Sciences,
46(17), 2621–2641.
[4] Boer, G. J. (1993). Climate change and the regulation of the surface moisture and energy budgets. Climate Dynamics, 8(5), 225–239.
http://dx.doi.org/10.1007/BF00198617
[5] Clapp, R. B., & Hornberger, G. M. (1978). Empirical equations for
some soil hydraulic properties. Water Resources Research, 14(4), 601–604.
http://dx.doi.org/10.1029/WR014i004p00601
[6] Collins, M., R., Knutti, J., Arblaster, J.-L., Dufresne, T. Fichefet, P. Friedlingstein,
X. Gao, W.J. Gutowski, T. Johns, G., & Krinner, M. Shongwe, C. Tebaldi, A. J.
W. and M. W. (2013). 2013: Long-term climate change: projections, commitments
and irreversibility. In T. F. Stocker, D. Qin, G.-K. Plattner, M. Tignor, S. K. Allen,
J. Boschung, . . . P. M. Midgley (Eds.), Climate Change 2013: The Physical Science
Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (pp. 1031–1106). United Kingdom and New
York, NY, USA.: Cambridge University Press.
[7] De Vries, D. A. (1975). Heat and mass transfer in the biosphere. In Heat transfer in
soils (pp. 5–28). Scripta.
[8] ECMWF. (2013). IFS Documentation - Cy40r1. Operational implementation 22
November 2013. Part IV: Physical processes. Retrieved December 1, 2014, from
http://www.ecmwf.int/sites/default/files/IFS_CY40R1_Part4.pdf
90
[9] European Supersites for Atmospheric Aerosol Research. (n.d.). Description of the SMEAR II station. Retrieved November 29, 2014, from
http://www.eusaar.net/upload/SMEAR2.pdf
[10] Farouki, O. T. (1986). Thermal properties of soils. Volume 11 of Series on Rock and
Soil Mechanics Series. Trans Tech Publications, Limited.
[11] Farquhar, G. D., von Caemmerer, S., & Berry, J. A. (1980). A biochemical model
of photosynthetic CO2 assimilation in leaves of C3 species. Planta, 149(1), 78–90.
http://dx.doi.org/10.1007/BF00386231
[12] Garratt, J. R. (1977). Review of drag coefficients over oceans and continents. Monthly Weather Review, 105(7), 915–929. http://dx.doi.org/10.1175/15200493(1977)105<0915:RODCOO>2.0.CO;2
[13] Garratt, J. R. (1994). The atmospheric boundary layer. Cambridge University Press.
[14] Ghuman, B. S., & Jalota, S. K. (2006). Heat capacity of soil. In R. Lal (Ed.), Encyclopedia of soil science (Vol. 2) (pp. 811–813). CRC Press.
[15] Goody, R. (2007). Maximum entropy production in climate theory. Journal of the
Atmospheric Sciences, 64(7), 2735–2739. http://dx.doi.org/10.1175/JAS3967.1
[16] Grassl, H. (1981). The climate at maximum entropy production by meridional atmospheric and oceanic heat fluxes. Quarterly Journal of the Royal Meteorological
Society, 5(1981), 153–166. http://doi.org/10.1002/qj.49710745110
[17] Greve, M. H., Helweg, A., Yli-Halla, M., Eklo, O. M., Nyborg, A. A., Solbakken, E.,
. . . Stenström, J. (1998). Nordic reference soils. (E. Tiberg, Ed.). Nordic Council of
Ministers.
[18] Grimmond, C. S. B., & Oke, T. R. (1995). Comparison of heat fluxes
from summertime observations in the suburbs of four North American cities.
Journal of Applied Meteorology, 34(4), 873–889. http://dx.doi.org/10.1175/15200450(1995)034<0873:COHFFS>2.0.CO;2
[19] Gutowski, W. J., Gutzler, D. S., & Wang, W. C. (1991). Surface energy balances of three general circulation models: Implications for simulating regional climate change. Journal of Climate, 4(2), 121–134. http://dx.doi.org/10.1175/15200442(1991)004<0121:SEBOTG>2.0.CO;2
[20] Hauf, T., & Höller, H. (1987). Entropy and potential temperature. Journal
of the Atmospheric Sciences, 44(20), 2887–2901. http://dx.doi.org/10.1175/15200469(1987)044<2887:EAPT>2.0.CO;2
91
[21] Held, I. M., & Soden, B. J. (2000). Water vapor feedback and global
warming. Annual Review of Energy and the Environment, 25(1), 441–475.
http://dx.doi.org/10.1146/annurev.energy.25.1.441
[22] Held, I. M., & Soden, B. J. (2006). Robust responses of the hydrological cycle to global warming. Journal of Climate, 19(21), 5686–5699.
http://dx.doi.org/10.1175/JCLI3990.1
[23] Henderson-Sellers, A., & Wilson, M. F. (1983). Surface albedo data
for climatic modeling. Reviews of Geophysics,
21(8),
1743–1778.
http://dx.doi.org/10.1029/RG021i008p01743
[24] Holton, J. R. (1973). An Introduction to Dynamic Meteorology. American Journal
of Physics. http://dx.doi.org/10.1119/1.1987371
[25] Ilvesniemi, H., Pumpanen, J., Duursma, R., Hari, P., Keronen, P., Kolari, P., . . .
Vesala, T. (2010). Water balance of a boreal Scots pine forest. Boreal Environment
Research, 15(4), 375–396.
[26] Inamdar, A. K., & Ramanathan, V. (1998). Tropical and global scale
interactions among water vapor, atmospheric greenhouse effect, and surface temperature. Journal of Geophysical Research, 103(D24), 32177–32194.
http://dx.doi.org/10.1029/1998jd900007
[27] IPCC. (2001). Climate change 2001: The scientific basis. (J. T. Houghton, D. J. Y.
Ding, M. Griggs, P. J. Noguer, X. van der Linden, K. Dai, . . . C. A. Johnson, Eds.)
Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change (Vol. 881). Cambridge: Cambridge University
Press.
[28] Jarvis, P. G. (1976). The interpretation of the variations in leaf water potential and stomatal conductance found in canopies in the field. Philosophical Transactions of the Royal Society B: Biological Sciences, 273(927), 593–610.
http://dx.doi.org/10.1098/rstb.1976.0035
[29] Kleidon, A. (2003). The atmospheric circulation and states of
imum entropy production. Geophysical Research Letters, 30(23),
http://dx.doi.org/10.1029/2003GL018363
[30] Kleidon,
A. (2010). Life,
hierarchy,
chinery of planet Earth. Physics of
http://dx.doi.org/10.1016/j.plrev.2010.10.002
and
Life
max1–4.
the thermodynamic maReviews, 7(4), 424–460.
[31] Kleidon, A. (2011). How does the earth system generate and maintain thermodynamic disequilibrium and what does it imply for the future of the planet?
92
Philosophical Transactions of the Royal Society of London, 370(1962), 1012–1040.
http://dx.doi.org/10.1098/rsta.2011.0316
[32] Kleidon, A. (2015). MPP limit in a radiative heat engine. Unpublished manuscript.
Jena.
[33] Kleidon, A., Fraedrich, K., Kirk, E., & Lunkeit, F. (2006). Maximum entropy production and the strength of boundary layer exchange in an atmospheric general circulation model. Geophysical Research Letters, 33(6), 2–5.
http://dx.doi.org/10.1029/2005GL025373
[34] Kleidon, A., & Renner, M. (2013). Thermodynamic limits of hydrologic cycling within
the Earth system: concepts, estimates and implications. Hydrology and Earth System
Sciences, 17(7), 2873–2892. http://dx.doi.org/10.5194/hess-17-2873-2013
[35] Kleidon, A., & Renner, M. (2014). Limits of land-atmosphere exchange. Unpublished
manuscript. Jena.
[36] Kleidon, A., Renner, M., & Porada, P. (2014). Estimates of the climatological land
surface energy and water balance derived from maximum convective power. Hydrology and Earth System Sciences, 18(6), 2201–2218. http://dx.doi.org/10.5194/hess18-2201-2014
[37] Konings, A. G., Feng, X., Molini, A., Manzoni, S., Vico, G., & Porporato, A. (2012).
Thermodynamics of an idealized hydrologic cycle. Water Resources Research, 48(5).
http://dx.doi.org/10.1029/2011WR011264
[38] Lacis, A. A., & Hansen, J. (1974). A parameterization for the absorption of solar
radiation in the Earth’s atmosphere. Journal of the Atmospheric Sciences, 31(1),
118–133. http://dx.doi.org/10.1175/1520-0469(1974)031<0118:APFTAO>2.0.CO;2
[39] Lorenz, R. D., Lunine, J. I., Withers, P. G., & McKay, C. P. (2001). Titan, Mars
and Earth: Entropy production by latitudinal heat transport. Geophysical Research
Letters, 28(3), 415–418. http://dx.doi.org/10.1029/2000GL012336
[40] Lovelock, J. E. (1975). Thermodynamics and the recognition of alien biospheres.
Proceedings of the Royal Society of London. Series B, Biological Sciences, 189(1095),
167–181.
[41] Manabe, S. (1969). Climate and the ocean circulation 1. The atmospheric circulation
and the hydrology of the Earth’s surface. Monthly Weather Review, 97(11), 739–774.
http://dx.doi.org/10.1175/1520-0493(1969)097<0739:CATOC>2.3.CO;2
93
[42] Moss, R. H., Edmonds, J. A., Hibbard, K. A., Manning, M. R., Rose, S. K.,
van Vuuren, D. P., . . . Wilbanks, T. J. (2010). The next generation of scenarios for climate change research and assessment. Nature, 463(7282), 747–756.
http://dx.doi.org/10.1038/nature08823
[43] Myhre, G., Highwood, E. J., Shine, K. P., & Stordal, F. (1998). New estimates of
radiative forcing due to well mixed greenhouse gases. Geophysical Research Letters,
25(14), 2715–2718. http://dx.doi.org/10.1029/98GL01908
[44] Myhre, G., Shindell, D., Bréon, F.-M., Collins, W., Fuglestvedt, J., Huang, J., . . .
Zhan, H. (2013). 2013: Anthropogenic and natural radiative forcing. In T. F. Stocker,
D. Qin, G.-K. Plattner, M. Tignor, S. K. Allen, J. Boschung, . . . P. M. Midgley
(Eds.), Climate Change 2013: The Physical Science Basis. Contribution of Working
Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate
Change (pp. 659–740). United Kingdom and New York, NY, USA.: Cambridge University Press.
[45] Nakajima, S., Hayashi, Y.-Y., & Abe, Y. (1992). A study on the “runaway greenhouse effect” with a one-dimensional radiative–convective equilibrium model. Journal of the Atmospheric Sciences, 49(23), 2256–2266. http://dx.doi.org/10.1175/15200469(1992)049<2256:ASOTGE>2.0.CO;2
[46] NCDC. (2015). Local climatological data for Barstow, California. Retrieved March
6, 2015, from http://www.ncdc.noaa.gov/cdo-web/datasets
[47] Ouwersloot, H. G., Vilà-Guerau De Arellano, J., Nàlscher, a. C., Krol, M. C.,
Ganzeveld, L. N., Breitenberger, C., . . . Lelieveld, J. (2012). Characterization of
a boreal convective boundary layer and its impact on atmospheric chemistry during
HUMPPA-COPEC-2010. Atmospheric Chemistry and Physics, 12(19), 9335–9353.
http://dx.doi.org/10.5194/acp-12-9335-2012
[48] Ozawa, H., Ohmura, A., Lorenz, R. D., & Pujol, T. (2003). The second law of thermodynamics and the global climate system: A review of
the maximum entropy production principle. Reviews of Geophysics, 41(4).
http://dx.doi.org/10.1029/2002RG000113
[49] Paltridge, G. W. (1979). Climate and thermodynamic systems of maximum dissipation. Nature, 279(5714), 630–631. http://dx.doi.org/10.1038/279630a0
[50] Ramanathan, V., & Vogelmann, A. M. (1997). Greenhouse effect, atmospheric solar
absorption and the Earth’s radiation budget: From the Arrhenius-Langley era to the
1990s. Ambio, 26(1), 38–46.
[51] Schmiedl, T., & Seifert, U. (2007). Efficiency at maximum power: An analytically
solvable model for stochastic heat engines. A Letters Journal Exploring the Frountiers
of Physics, 81(2), 1–6. http://dx.doi.org/10.1209/0295-5075/81/20003
94
[52] Staley, D. O., & Jurica, G. M. (1972). Effective atmospheric emissivity under clear skies. Journal of Applied Meteorology, 11(2), 349–356.
http://dx.doi.org/10.1175/1520-0450(1972)011<0349:EAEUCS>2.0.CO;2
[53] Stull, R. B. (1988). An introduction to boundary layer meteorology (Vol. 13). Springer
Netherland. http://doi.org/10.1007/978-94-009-3027-8
[54] Suni, T., Rinne, J., Reissell, A., Altimir, N., Keronen, P., Rannik, U., . . . Vesala,
T. (2003). Long-term measurements of surface fluxes above a Scots pine forest in
Hyytiala, southern Finland, 1996-2001. Boreal Environment Research, 8(4), 287–301.
[55] Trenberth, K. E., Fasullo, J. T., & Kiehl, J. (2009). Earth’s global energy budget. Bulletin of the American Meteorological Society, 90(3), 311–323.
http://dx.doi.org/10.1175/2008BAMS2634.1